Author: Mona R.

Microsoft announced on June 12 that it is eliminating approximately 2,000 positions across Xbox Game Studios, Activision Publishing, and Blizzard Entertainment. The cuts represent roughly 8 percent of the combined gaming headcount Microsoft inherited when it closed its $68.7 billion acquisition of Activision Blizzard in October 2023. Phil Spencer, head of Microsoft Gaming, framed the announcement in terms of AI-assisted development tools that are, in his words, fundamentally changing how Microsoft creates games at scale. The workforce reduction, he said in an internal memo, would allow teams to do more with the focused resources the company is bringing forward.

That framing deserves close reading. Microsoft is not claiming the business declined and therefore needs fewer people. It is claiming the business can produce the same or better output with fewer people because AI tools now fill roles that humans previously occupied. That is a different argument — with different implications for the employees affected, for Microsoft’s financial position, and for what the game development sector can expect over the next five years.

The June cuts did not happen in isolation. The gaming sector shed approximately 10,000 jobs in 2024 across EA, Sony Santa Monica, Unity, Bungie, and dozens of smaller studios — a wave that industry analysts attributed to post-pandemic demand correction combined with interest-rate-driven cost pressure. Microsoft itself contributed to that 2024 wave with the layoff of approximately 1,900 Xbox and Activision employees in January 2024, followed by the closure of Tango Gameworks, Arkane Austin, and Alpha Dog Games in May 2024. What makes the June 2026 round different is not its scale but its stated justification. Microsoft is the first major game publisher to cite AI tool deployment as the primary driver of a large involuntary workforce reduction — not demand normalization, not portfolio rationalization, but technology replacement. That distinction extends the implications of this announcement well beyond the gaming sector.

The gaming sector is not the first creative industry to face this argument. Music labels in the streaming era, visual effects houses in the AI compositing era, and news organizations in the algorithmic curation era all experienced versions of the same restructuring claim: technology enables the same output with fewer people. The game development test is different in one important way — the output is interactive, iterative, and quality-sensitive across thousands of hours of player experience in ways that algorithmic music recommendation or AI-assisted compositing are not. The proof standard is high, and it is public and observable.

The Xbox layoffs are the first large-scale test, at a public company with measurable output metrics, of whether AI productivity tools can replace a meaningful portion of a creative and technical workforce without visible degradation in product quality. The answer will arrive over the next two to three years in the form of shipping games, review scores, and Game Pass subscriber retention. If the bet works, it changes the calculus on AI workforce displacement across the broader technology industry. If it does not, it is the most visible public counter-evidence to date against the AI productivity thesis that Microsoft’s $190 billion capex position requires to justify itself.

What the Voluntary Buyout Did Not Achieve

The June cuts did not arrive without warning. In Q1 2026, Microsoft offered a voluntary departure package to Xbox and Activision employees across its gaming divisions. Internal communications reviewed by industry outlets indicated Microsoft expected between 60 and 70 percent of eligible senior roles to participate, which would have achieved its restructuring targets without forced separations.

Fewer than half of eligible employees accepted the package. The undersubscription forced Microsoft into the position it publicly committed to avoiding after the 2023 acquisition: involuntary layoffs within the Activision organization during a period of cultural integration. The voluntary buyout was already described as addressing only 7 percent of the structural problem facing the combined organization — a reorganization that needed to rationalize duplicated functions across the Microsoft, Xbox, Activision, and Blizzard layers without triggering the regulatory and reputational scrutiny that forced cuts generate. The low voluntary uptake partly reflects the lesson employees drew from the 2024 studio closures: that Microsoft’s restructuring decisions are not performance-contingent and are not reversed by employee cooperation with voluntary programs.

When the voluntary approach fell short, the forced cuts became necessary. The 2,000 number represents approximately the gap between what voluntary departures achieved and what Microsoft’s restructuring model required to reach its target cost structure for the gaming division.

Where the 2,000 Jobs Were

The cuts are concentrated in three functional areas: quality assurance testing, publishing operations, and consumer marketing. These are not the roles most visible in game credits, but they represent a substantial portion of any large studio’s total workforce and a disproportionate share of the cost structure at Activision-scale operations.

Quality assurance at scale is labor-intensive. Large titles at Activision and Blizzard run test teams of several hundred people cycling through regression testing, console platform compliance, localization verification, and accessibility certification across multiple regions. At peak production on a Call of Duty title, the QA footprint has historically run 400 to 600 testers across three time zones — a workforce structure designed for human throughput on known test cases rather than automated coverage of dynamic game states.

Microsoft has been deploying AI-assisted QA tooling across its studios since early 2026, claiming that automated test generation and failure identification can cover 60 to 80 percent of regression testing volume that previously required human testers. The tooling generates test scripts from game build changes, identifies regressions by comparing output states against prior validated builds, and flags failures with enough specificity that human testers can investigate root causes rather than run the full case suite manually. If the 60 to 80 percent coverage claim holds in production, it justifies a meaningful reduction in QA headcount per title on regression-heavy test phases. The remaining 20 to 40 percent — complex interaction testing, subjective quality assessment, performance profiling on edge-case hardware configurations — remains human work that automated systems cannot yet reliably handle.

Publishing operations have been consolidating since Game Pass shifted the majority of Xbox first-party releases to a subscription model. When a title launches day-one on Game Pass rather than through retail-primary distribution, the publishing workflow requires fewer coordination roles between developer, platform holder, and retailer. Consumer marketing has been rationalized toward platform-level subscription marketing rather than title-by-title campaign staffing. The geography of the cuts reflects these functional concentrations. Activision’s Call of Duty mobile team in Austin lost approximately 400 positions. Blizzard’s licensing and consumer products team in Irvine lost approximately 300. Microsoft Game Studios in Redmond eliminated approximately 400 roles in QA automation and test infrastructure. The remaining 900 cuts spread across publishing and marketing functions at multiple studio locations.

The Union Question Microsoft Will Have to Answer

The game industry’s labor organization has changed substantially since Microsoft acquired Activision. Raven Software’s QA workers voted to form the first recognized union at a major US game studio in May 2022, and Microsoft committed at acquisition close to recognizing the union and negotiating in good faith. Several additional organizing drives have succeeded across the Xbox and Activision organization in the years since. The June 2026 cuts include positions at studios where collective bargaining agreements are now in effect.

The AI replacement justification creates a specific legal and reputational challenge for those union relationships. Standard workforce reduction provisions in labor agreements typically distinguish between economic layoffs — where the business can no longer afford the roles — and technological displacement, where the business replaces human roles with automation. The obligations attached to each can differ significantly: longer notice periods, retraining rights, preferential rehire rights for roles subsequently re-opened, and in some agreements, requirements to bargain over the decision to automate before implementation rather than simply bargain over its effects.

Microsoft has not publicly clarified how its AI-displacement rationale intersects with its collective bargaining obligations. If the cuts at unionized studios are classified as economic layoffs rather than technological displacement, the union agreements may permit them under standard reduction-in-force provisions. If they are classified as technological displacement — which the AI framing implies by being explicit that tools are replacing roles — affected union members and their representatives have grounds to request bargaining over the automation decision before it takes effect. That question has not been resolved in public disclosures, and it represents a legal and reputational exposure that the AI justification creates specifically by being so explicit about technology replacement as the driver.

This is the first time a major employer has stated so clearly that AI tools are why specific roles are being eliminated. Future workforce reductions in technology and media that cite AI replacement will be measured against how Microsoft handles the obligations its own AI framing creates. The question will be watched closely by the broader game industry’s organized labor community, which has been adding bargaining units at a pace that would have been unthinkable five years ago. The outcome here sets a precedent that every subsequent AI-replacement layoff in the sector will be measured against.

The 2024 Pattern and What It Established

The June 2026 cuts follow a pattern the 2024 studio closures established. In May 2024, Microsoft announced the closure of Tango Gameworks — the studio that had just released Hi-Fi Rush, a title Microsoft itself described as a commercial and critical success — alongside Arkane Austin. Those closures were not framed in AI terms. They were framed as portfolio rationalization: Microsoft had more studios than its resource allocation model could productively support.

The 2024 closures established that Microsoft’s commitment to any individual studio or franchise is contingent on portfolio-level decisions, not on the commercial or critical performance of individual titles. That precedent informed how employees across the Xbox organization read the voluntary buyout offer in Q1 2026. The offer was read less as a generous exit and more as advance notice that restructuring was coming regardless, with the voluntary terms potentially better than what would follow. That reading explains the low acceptance rate and helps account for why Microsoft ended up in the forced-layoff position it had said it wanted to avoid.

The 2024 pattern also established something about how Microsoft thinks about the Activision acquisition’s asset base. The studios it has closed were small, creatively independent teams making games that did not fit the Game Pass subscriber acquisition model at the cost required to make them. Closing them was not a declaration that Microsoft does not believe in first-party game development. It was a declaration that Microsoft believes in first-party game development specifically through the lens of what drives Game Pass subscriber acquisition and retention at scale. The June 2026 workforce reduction is consistent with that lens: the functions being eliminated are those least directly tied to the creative output that Game Pass subscribers are paying for.

The Xbox Hardware Context

The workforce reduction arrives as Xbox hardware revenue declines for a third consecutive quarter. The Series X and Series S have not recaptured the unit sales trajectory Microsoft projected when pricing the Activision acquisition. Game Pass subscriber growth has continued at a pace that, in isolation, would be considered strong for any subscription media service, reaching approximately 45 million in the most recent quarterly disclosure. But the subscriber acquisition cost — including content investment required to drive subscriptions and the amortized cost of the Activision library now included in the service — has compressed unit economics relative to the projections that underwrote the deal. Against this backdrop, the workforce reduction is a margin move alongside a genuine AI tooling transition. Smaller workforces cost less regardless of whether AI tools replace the eliminated roles fully, partially, or not at all.

The Counterargument: Microsoft Is Still Investing in Games

The bear read — that Microsoft is retreating from first-party game development and treating Xbox primarily as a subscription delivery mechanism — requires confronting evidence that runs against it. Microsoft has not announced studio closures alongside the June workforce reduction. The cuts are distributed across support functions, not concentrated in the elimination of specific creative teams. The studios responsible for Microsoft’s most commercially critical franchises — 343 Industries for Halo, The Coalition for Gears of War, Infinity Ward for Call of Duty — have not seen announced layoffs in this round. Microsoft is also continuing to invest in first-party development pipelines with no announced changes to active release schedules. On this reading, the June cuts are a rationalization of duplicated support infrastructure — not a retreat from the game development function.

Why the AI Productivity Bet Has Not Yet Been Proven at This Scale

The problem with the productivity argument is that it requires the AI capability claim to hold at the creative and production quality that competitive commercial game development demands — and that has not yet been demonstrated at the scale Microsoft is now betting on.

QA automation and localization tools are proven in their core applications. The claim that AI-assisted teams can produce competitive titles with 30 to 40 percent fewer people across creative and coordination functions has not been proven. It is a prediction about tools in early deployment whose real test requires shipping titles and measuring their quality against prior releases made by larger teams. The development timelines mean the evidence will not arrive before 2028 at the earliest for titles currently in early production under the new model.

The financial stake attached to proving this claim is significant. The Activision acquisition closed at $68.7 billion, a multiple that required a specific thesis about the gaming market’s future to justify. If the Xbox AI productivity test fails — if the games produced by smaller AI-assisted teams are materially lower quality, or if development timelines stretch rather than compress — it damages the credibility of the AI productivity thesis at exactly the moment when Microsoft needs that thesis vindicated at the enterprise level, where Copilot at 3.3 percent enterprise penetration has not yet provided the vindication the $190 billion capex requires. Xbox becomes the internal test of the external claim.

The broader enterprise AI adoption data suggests the risk is real. AI tools have demonstrated productivity gains in narrow, discrete, measurable tasks. They have not demonstrated the ability to replace creative and coordination functions in complex production workflows at the quality level that competitive commercial products require. Game development — combining creative, technical, and high-stakes production-coordination demands across multi-year timelines — is exactly the kind of complex workflow where horizontal AI productivity tools have been weakest. Microsoft is betting that gaming is the exception. The bet is now in production.

Every subsequent evaluation of Microsoft’s gaming strategy will be made against the output of studios that absorbed these headcount reductions. Subscription renewal rates, average review scores, post-launch patch volumes as a proxy for QA coverage quality, and development timelines will all function as measurable indicators of whether the AI productivity claim holds in practice. The workforce that remains at Xbox Game Studios after the June cuts is smaller, more AI-tool-dependent, and facing a more visible performance test than any cohort of game developers in Microsoft’s history. Whether they meet it will count for more than any benchmark test or analyst estimate in the current AI investment cycle — because the results will be public, observable, and commercially significant at a scale that internal productivity metrics never are.

The Record: What Microsoft Said About Developers Before Each Round of Cuts

Accountability journalism requires a timeline. The pattern in Microsoft gaming statements over the past three years is not one of sudden strategic reversals. It is one of incremental commitment followed by incremental withdrawal, each transition accompanied by language that frames the contraction as forward investment rather than retreat. Documenting the record is the precondition for evaluating whether the current language is a better predictor of the current outcome than the previous language was of the previous one.

The January 2024 layoffs of 1,900 people across Activision Blizzard, Bethesda, and Xbox studios were announced alongside statements about streamlining operations for the next generation of gaming experiences and investing in capabilities that would define Xbox’s future. Eighteen months later, 2,000 additional people are being separated through voluntary buyouts and strategic realignment. The language evolves. The employment count moves in one direction. The accountability question is whether the analysis that would justify the 2025 round, if applied in early 2024, would have predicted the 2024 round. The answer to that question is what determines whether the current round is the last one or the latest one.

The AI productivity framing is the claim that most requires scrutiny. Microsoft has stated consistently that AI tools allow smaller developer teams to produce games that previously required larger ones. That is a testable prediction. It requires that the titles produced by AI-assisted smaller teams are equivalent in quality and commercial performance to the titles produced by the larger teams they replaced. No such comparison data has been published. The claim is being used to justify workforce reduction before the productivity gain has been validated at the scale the reduction implies.

The enterprise AI adoption baseline is relevant here as a cross-industry reference point. Across industries, AI productivity tools are demonstrating meaningful gains on specific task types: code completion, asset generation, testing automation. But team-level output gains are significantly more modest and slower to materialize than task-level gains imply. A developer who writes code 30% faster with AI assistance does not produce a game 30% faster, because game development bottlenecks are not primarily in code generation speed. They are in design iteration, quality assurance, cross-functional coordination, and narrative development, domains where AI productivity gains are less established and harder to measure.

The Microsoft developer platform dynamic adds a structural context that the gaming-specific framing misses. Microsoft has been systematically increasing the margin it extracts from its developer tooling while reducing the internal developer headcount that uses those tools. That is not incoherent as a strategy: if AI tools genuinely increase developer productivity at the team output level, you need fewer developers to produce the same output. But the strategic logic requires that the productivity claim be true at the team output level, not just at the individual task level. The gap between individual task productivity and team output productivity is where most corporate AI productivity claims have overestimated the near-term benefit.

The union question identified in the article is the mechanism that would make the accountability question empirically legible. If the Communication Workers of America establishes representation at ZeniMax and that representation includes transparency provisions on AI tool deployment and workforce levels, the productivity claim becomes auditable rather than asserted. An agreement requiring Microsoft to report on the relationship between AI tool adoption rates and headcount changes would make the AI-productivity rationale testable. Without that data, the external observer cannot distinguish between AI-driven productivity gain and budget-target-driven reduction with AI productivity cited as the rationale.

The hidden cost of large-scale gaming layoffs is institutional knowledge loss. Friction is the silent cost driver in any knowledge-intensive organization, and game development is among the most knowledge-intensive production processes in the entertainment industry. The developers who know a specific engine’s edge cases, a specific franchise’s player community expectations, and a specific game’s history of design decisions carry knowledge that does not transfer through documentation. When those people leave, the friction cost appears in the next production cycle as longer timelines, more quality issues at launch, and less accuracy in predicting what the audience will value. That cost does not appear in the quarterly earnings where the labor savings appear. It appears in the title performance data two to three years later.

The Chinese AI competitive development creates genuine pressure on Western game publishers to demonstrate AI productivity gains. Chinese gaming companies are deploying AI-assisted development tools aggressively and may reach production-scale AI-generated content faster than Western publishers in specific categories. That competitive pressure is real. But the rational response to competitive AI deployment pressure is to use AI tools to make experienced developers more productive, not to reduce the experienced developer base before the tools have demonstrated their productivity claim at scale. The optimized response to the competitive threat is the combination. The current response appears to be betting on the AI tools to substitute for the combination.

Prediction markets on Microsoft gaming revenue are pricing modest growth over the next three years, consistent with Game Pass subscriber growth continuing at a decelerating rate partially offset by AI-assisted cost reduction. What those markets are not pricing is the tail scenario where AI productivity gains fail to materialize at the team output level and institutional knowledge loss from successive layoffs surfaces in title performance by 2027-2028. That scenario is not yet legible in the market price because the first post-reduction titles have not shipped. The accountability timeline will provide the data. The question is whether the organization will still have the institutional knowledge to interpret it correctly when it arrives.

Defense stocks have been the quiet sector outperformer of the past three years across both US and European markets. Lockheed Martin, Northrop Grumman, RTX, General Dynamics, and the smaller US prime contractors have produced total returns that have outpaced most cyclical sectors. The European defense companies — BAE Systems, Rheinmetall, Saab, Leonardo, Hensoldt, and several specialised manufacturers — have produced even more dramatic outperformance, with several of these stocks delivering multi-hundred-percent returns since the 2022 inflection point that began the European rearmament cycle.

The structural drivers of the outperformance are genuinely durable in ways that the market continues to underprice. The European NATO commitment to defense spending at significantly elevated levels, the US defense budget trajectory under both political parties, and the global geopolitical environment that has hardened across multiple flashpoints all support a procurement cycle that operates over multi-year and multi-decade time horizons rather than the cyclical quarters that typically dominate equity market positioning.

Understanding why the defense sector outperformance has been sustained, what the specific procurement dynamics actually look like, and which companies have the strongest competitive positions for the continuation of the cycle provides important context for evaluating both the defense sector exposure and the broader implications of sustained geopolitical reality for global investment allocation.

The European Rearmament Commitment That Has Not Been Fully Discounted

The European NATO members’ commitment to defense spending at 2 percent of GDP and beyond — and the specific spending plans that several governments have announced for periods well above the 2 percent threshold — represents a structural change in the European defense procurement environment that the market has been slow to fully discount. The aggregate increase in European defense spending from the pre-2022 baseline to the current commitment levels represents hundreds of billions of euros in incremental procurement over the next decade, supporting sustained revenue growth for the manufacturers who can satisfy the demand.

The specific procurement programs that have been initiated or expanded include the various air defense systems (Patriot extensions, IRIS-T expansion, the European Sky Shield Initiative), the artillery and ammunition production capacity expansion (driven significantly by the lessons of the Ukraine conflict about munitions consumption rates in sustained conventional warfare), the ground combat vehicle programs across multiple European nations, and the broader military aircraft, naval, and electronic warfare procurement that operates on multi-year and multi-decade timelines.

The broader European equity outperformance has been substantially supported by the defense sector contribution, and the defense sector dynamics deserve specific consideration separate from the broader European macro analysis. The defense procurement is driven by political commitment rather than by the cyclical economic dynamics that affect most other sectors, which means the defense sector revenue visibility is genuinely different from typical industrial cyclical dynamics.

The specific European defense beneficiaries have been varied in their performance based on their product positioning. Rheinmetall has captured significant value as the dominant European producer of ammunition and ground combat systems. Saab has benefited from increased fighter aircraft and submarine demand. Leonardo has captured value across helicopters, electronics, and various other defense categories. BAE Systems has benefited from sustained UK and international demand. Hensoldt has been particularly successful in the electronic warfare and sensor systems segment that has received elevated funding given the modern combat environment’s emphasis on electromagnetic spectrum capability.

The US Defense Budget and Procurement Reality

The US defense budget continues to operate at substantially elevated levels in real terms, supported by both political parties’ general commitment to defense spending despite political differences on specific priorities. The annual defense budget exceeds $850 billion in the current cycle, with the trajectory continuing to grow modestly in nominal terms even as inflation considerations affect the real growth rate.

The specific US defense procurement priorities that affect the major contractors include the various nuclear modernisation programs (replacing the Cold War-era nuclear delivery systems), the F-35 program that continues to support Lockheed Martin’s revenue, the various unmanned systems and AI-related defense modernisation, the shipbuilding programs that support General Dynamics and Huntington Ingalls, and the broader missile and air defense procurement that affects RTX, Lockheed, and the various other prime contractors.

The honest assessment of the US defense procurement environment is that the cycle is structurally supportive of the major contractors but has more specific competitive dynamics than the European cycle. The US procurement process produces specific winners and losers among the prime contractors based on program awards, performance, and political dynamics that affect specific programs. The aggregate US defense contractor exposure provides reasonable broad sector returns; the specific contractor selection requires more careful evaluation of program positioning.

The defense technology and modernisation themes have produced substantial growth for the companies that have positioned for these segments. The broader AI infrastructure investment has affected defense modernisation in ways that benefit specific technology-adjacent defense companies (Palantir for software and AI, Anduril for autonomous systems, the various other defense tech companies that have integrated modern computing capabilities into traditional defense applications).

The Procurement Cycle Duration and Revenue Visibility

The structural feature of defense procurement that distinguishes it from most other industrial sectors is the duration of the procurement cycles and the resulting revenue visibility. A modern military procurement program from initial requirement definition through delivery and sustainment can span 20-30 years. A fighter aircraft program operates over similar timelines. Submarine and naval shipbuilding programs operate over 30-50 year timelines including initial construction and the sustainment activity throughout the platform lifecycle.

This procurement cycle duration provides revenue visibility that is genuinely different from cyclical industrial demand. A defense contractor with substantial backlog in major platform programs has revenue visibility that extends years into the future, with limited risk of cancellation given the political and operational commitment that the programs represent. The combination of multi-year backlogs and the political durability of major defense programs creates an income stream that is more stable than most industrial revenue.

The book-to-bill ratios for the major defense contractors have been elevated for sustained periods since 2022, indicating that new orders are exceeding revenue recognition by meaningful margins. This is consistent with sustained backlog growth that will support revenue growth in future years. The specific book-to-bill data for the major US and European contractors has been monitored closely by analysts as a leading indicator of the defense sector revenue trajectory.

The Ammunition and Munitions Sub-Sector

The ammunition and munitions sub-sector deserves specific attention because it has been the most acute and visible expression of the post-2022 defense procurement dynamics. The consumption rates for artillery ammunition, anti-tank weapons, and various other munitions in sustained conventional warfare have substantially exceeded pre-2022 production capacity assumptions, requiring major industrial capacity expansion across the Western defense industrial base.

The specific companies that have benefited from the munitions production expansion include Rheinmetall (artillery ammunition and Leopard tank ammunition), General Dynamics (ammunition production capacity in the US), Northrop Grumman (various missile and munitions systems), Nammo (the Nordic explosives and ammunition manufacturer), and several smaller specialised manufacturers that have received substantial orders to expand their production capacity.

The capacity expansion that the munitions sub-sector has executed represents long-term commitments that provide sustained revenue beyond the immediate Ukraine-driven demand. The political consensus across Western nations about the importance of munitions production capacity (driven by the lessons of conventional warfare consumption rates) supports continued demand even if the specific Ukraine-related demand moderates. The munitions sub-sector therefore has structural support for sustained elevated production through the procurement cycle.

The Software and AI Defense Categories

The intersection of defense procurement with AI and software capabilities has created specific company opportunities that operate differently from the traditional defense prime contractor model. Palantir Technologies has positioned its data analytics and AI capabilities for defense and intelligence applications, with substantial growth in revenue from defense customers. Anduril Industries has built autonomous systems and AI-enabled defense applications that have captured significant contracts from US and allied defense customers.

The strategic question for these defense technology companies is whether they can sustain the rapid growth that the early AI defense procurement has produced as the procurement cycle matures and as the traditional prime contractors invest in matching capabilities. The bull case is that the defense technology companies have structural advantages in pace of innovation, ability to recruit AI talent, and integration with broader commercial AI capabilities that the traditional primes cannot match. The bear case is that the traditional primes have substantially deeper political and procurement relationships that allow them to capture the AI-related procurement value as their offerings mature.

The probable outcome is some combination of both — the defense technology companies will sustain meaningful positions in specific niches where their AI and software capabilities provide clear differentiation, while the traditional primes will increasingly integrate AI capabilities into their broader programs and capture the value from the broader defense AI procurement. The investment implications depend on identifying which specific companies have the most durable positions in their respective niches.

The Risks That Could Disrupt the Cycle

The defense sector outperformance has been sustained for long enough that the specific risks that could disrupt the cycle deserve consideration. The most significant near-term risk is political — the possibility of a shift in defense spending priorities across major Western nations that would reduce the procurement commitment levels that the current outperformance is based on. The political consensus across Western nations about the importance of elevated defense spending has been more durable than typical political coalitions, but it is not immune to changes in political circumstances.

The fiscal pressure that elevated defense spending creates is a real consideration. Defense spending at the levels that European NATO members have committed to consumes substantial fiscal space that could otherwise support other government priorities, and the political sustainability of these commitments depends partly on the public’s continued perception that the elevated spending is justified by the security environment. A scenario where geopolitical tensions ease meaningfully could produce political pressure to reduce defense commitments, with corresponding implications for the defense sector revenue trajectory.

The broader fiscal pressure on Western governments creates a complex dynamic where defense spending is one of several competing priorities for limited fiscal capacity. The political durability of defense spending has been demonstrated through the current cycle, but the structural fiscal pressure could eventually constrain the procurement growth that the current trajectories imply.

The Valuation and Positioning Considerations

Defense sector valuations have expanded substantially over the past three years reflecting the strong fundamental performance and the increased market recognition of the structural drivers. The current valuations are not at the depressed levels that supported the early outperformance, which means the marginal return from new positioning is more dependent on the procurement cycle continuing than on the multiple expansion that supported earlier returns.

For investors evaluating defense sector exposure in 2026: the structural case for continued exposure remains strong, but the entry valuations matter more than they did in 2022-2023. Selective positioning across the defense sector — emphasising companies with the strongest specific program positions, the most durable competitive advantages, and reasonable valuations relative to growth — produces better risk-adjusted returns than broad sector exposure at current valuations.

The European defense exposure continues to provide stronger relative value than US defense exposure because the European procurement cycle is at an earlier stage of its development relative to the US cycle. The specific European companies that have positioned for the rearmament cycle have produced substantial returns but continue to trade at valuations that reflect their growth opportunity, while US defense valuations more fully reflect the established procurement environment.

For broader portfolio considerations: defense exposure provides specific characteristics — low correlation to broader cyclical equity dynamics, durable revenue visibility, and exposure to the geopolitical themes that continue to shape the global investment environment — that justify dedicated allocation rather than treating defense as an undifferentiated industrial sub-sector. The persistence of the defense sector outperformance has demonstrated that the structural drivers are real and durable, and the appropriate response is portfolio-level recognition of these drivers rather than continued underweighting of a sector that has produced sustained alpha for three years.

The honest position is that the defense sector outperformance reflects genuine structural drivers that continue to support continued investment exposure, that the entry valuations require more careful selection than they did at the cycle’s start, and that the multi-decade procurement visibility provides revenue durability that justifies premium multiples relative to typical industrial sectors. The next several years will continue to test the political commitment to elevated defense spending, but the trajectory of the current cycle suggests continued strong fundamental performance even as the valuation expansion that supported earlier returns moderates.

Testing Defense Stocks Against the Seven Powers Framework

Hamilton Helmer’s Seven Powers framework identifies the competitive advantages that generate durable, compounding returns rather than temporary outperformance. Defense stocks have delivered three years of exceptional returns, and the standard explanation — geopolitical tension, European rearmament, procurement backlogs — is accurate but incomplete. The Seven Powers analysis identifies which of those drivers represent genuine competitive moats and which represent cyclical tailwinds that will eventually normalize. The distinction matters for whether defense returns over the next five years look like the last three or like the mean-reversion that follows most cyclical outperformance.

The strongest defense sector power is Switching Costs, and it operates at a scale that makes most software switching costs look trivial. A nation-state that has integrated Lockheed’s F-35 into its air force cannot switch to a competitor’s platform without retiring the entire training pipeline, maintenance infrastructure, logistics system, spare parts supply chain, and pilot certification program that the F-35 requires. The switching cost of changing a major defense platform is measured in decades and billions of dollars. That is not a moat. That is a geological feature. The multi-decade revenue visibility the article identifies is a direct consequence of this Switching Cost structure: once a platform is adopted, the revenue stream that follows is close to captive.

Cornered Resource is the second relevant power, and it operates through the cleared workforce and classified program access that defines the prime contractors’ competitive position. Lockheed, Northrop, Raytheon, and BAE have accumulated decades of security clearances, classified program experience, and government customer relationships that cannot be quickly replicated by new entrants. A startup that builds better drone technology cannot access the classified threat assessment data that defines the performance requirements for the defense programs being competed. The Cornered Resource is not the technology. It is the security infrastructure, the clearance pipeline, and the classified customer relationships that determine who gets to compete for the programs that generate the revenue.

Counter-Positioning is where defense’s competitive landscape becomes most interesting relative to the AI technology sector. Enterprise AI adoption is creating a new category of defense-relevant capability — autonomous systems, AI-assisted intelligence analysis, cyber offense and defense — that the legacy prime contractors are not optimally positioned to develop. The new entrants in defense AI — Palantir, Anduril, Shield AI — are attempting a Counter-Positioning move: building AI-native defense capabilities that the legacy primes cannot replicate without cannibalizing their existing program structures. Whether that Counter-Positioning succeeds depends on whether the procurement system changes fast enough to allow AI-native capabilities to compete on equal terms with legacy platform contracts.

The European rearmament commitment is a genuine multi-year demand catalyst that is not fully reflected in the current order books. Rheinmetall and BAE’s ground vehicle and ammunition backlogs extend years into the future at production rates that are already being expanded. The corporate capital return context is relevant: defense companies are returning capital at rates that imply confidence in the demand visibility — they are not hoarding cash for uncertain times but distributing it because the procurement contracts make revenue sufficiently visible to support distributions. That is the behavioral signal from the management teams closest to the actual backlog data.

The Seven Powers risk in defense that the three-year return does not yet reflect is Network Economies — or more precisely, the absence of them in the legacy platform model. Software businesses generate network effects that compound returns as the user base grows. Defense platforms do not. The F-35’s value does not increase because more nations operate it; the contract is the unit of revenue, and the revenue is fixed by the contract rather than by a growing network. That means the defense sector’s returns are durable but not compounding — they grow at procurement cycle rates rather than at network economy rates. Prediction markets on European NATO defense spending through 2027 are pricing continued demand growth — which means the cyclical tailwind still has runway. But the Seven Powers analysis would price that tailwind at a modest multiple, not at a software-style compounding multiple. The sector is genuinely strong. It is not structurally exceptional in the way the recent returns imply.

Berachain launched its mainnet in early 2025 with one of the most ambitious tokenomic designs in recent Layer 1 history. The Proof-of-Liquidity (PoL) consensus mechanism, the three-token system (BERA for gas and value capture, BGT for governance and emissions, HONEY as the native stablecoin), and the explicit positioning of liquidity provision as the foundation of network security represented a genuine attempt to solve the cold-start liquidity problem that has constrained most new Layer 1 launches.

The early evidence about how the Berachain experiment has actually performed is now available for analysis. The mainnet has been operational for over a year, the BERA and BGT tokens have established trading patterns, the DeFi ecosystem has built out around the Proof-of-Liquidity incentive structure, and the broader competitive position relative to other Layer 1 challengers can be assessed with empirical data rather than just whitepaper projections.

Understanding what Berachain has actually built, what the Proof-of-Liquidity mechanism does in practice, and where the structural sustainability questions sit requires looking at the specific mechanics, the early ecosystem data, and the broader competitive context that Berachain operates within. The honest assessment includes both the genuine innovations that the protocol has demonstrated and the legitimate questions about whether the tokenomic structure can sustain through changing market conditions.

How Proof-of-Liquidity Actually Works

The Proof-of-Liquidity consensus mechanism is the central architectural innovation of Berachain. The system separates the validator security function from the liquidity provision function in a way that aims to align both with the broader network security and the application ecosystem development.

The mechanism works roughly as follows: validators stake BERA to participate in consensus, but the rewards that validators earn are paid in BGT (the governance token) rather than in BERA itself. Validators can direct the BGT rewards they earn to specific reward gauges (associated with specific DeFi protocols and liquidity pools) where the BGT flows to the liquidity providers in those pools. This creates an incentive structure where validators are economically incentivised to direct rewards to the gauges that have the most BGT bribes (payments from protocols seeking BGT emissions to their pools), which produces market-driven liquidity allocation across the ecosystem.

The HONEY stablecoin operates as the native dollar-pegged unit within the ecosystem, with various backing arrangements that include other crypto assets and integrations with broader stablecoin liquidity. HONEY is used in many of the DeFi applications on Berachain and provides the dollar unit that liquidity providers and traders use for activities within the ecosystem.

The architectural logic is that Proof-of-Liquidity aligns three interests that other consensus mechanisms keep separate: validator economics (BERA staking rewards), liquidity provider economics (BGT emissions to liquidity pools), and protocol ecosystem development (the bribe market that determines which protocols receive emissions). The hope is that this alignment produces sustained ecosystem development because the rewards distribution naturally flows to the protocols and pools that generate the most economic activity rather than to passively-held validator stakes.

The Early Ecosystem Development

The Berachain ecosystem development since mainnet launch has produced meaningful activity. The DeFi protocols that have launched on Berachain include various lending platforms, decentralised exchanges, and stablecoin issuers that have integrated with the Proof-of-Liquidity mechanism through bribe markets and BGT emissions targeting. The total value locked has grown to multiple billion dollars across the various protocols, supported partly by the BGT emissions and partly by the organic activity that the ecosystem has generated.

The specific protocols that have established meaningful positions in the Berachain ecosystem include BeraSwap (the major DEX), various lending protocols, and the broader infrastructure that supports DeFi activity on the chain. The ecosystem has been particularly active in stablecoin-related applications, with HONEY adoption supported by both the native protocol integration and by the broader ecosystem’s adoption of HONEY as a payment and trading unit.

The user activity metrics for Berachain have been reasonable for a Layer 1 in its first year of mainnet operation. Daily active addresses, transaction volumes, and the various engagement metrics have shown growth that is consistent with the kind of activity that BGT emissions would incentivise. The challenge is distinguishing between activity that is genuine economic activity and activity that is primarily about capturing BGT emissions — a distinction that affects how the ecosystem development should be interpreted.

The Tokenomic Sustainability Question

The honest critical evaluation of Berachain’s tokenomic structure has to confront the central question: whether the BGT emissions that drive much of the early ecosystem activity can be sustained at levels that support continued ecosystem development without producing the token economic dynamics that have undermined other emission-heavy protocols.

The pattern that emission-heavy protocols have historically followed is that the initial activity supported by emissions creates ecosystem development and user engagement, but the emissions themselves create selling pressure on the token as recipients of emissions sell to realise their economic gains. If the underlying ecosystem activity does not produce sufficient organic demand for the token to offset the emission-driven supply, the token price declines, which reduces the economic value of future emissions, which then reduces the incentive for liquidity providers to participate, which can create the negative feedback loop that has affected various other emission-driven protocols.

The Berachain team has designed mechanisms to address these concerns. The bribe market structure creates ongoing demand for BGT from protocols seeking emissions, the validator economics create demand for BERA from staking activity, and the HONEY stablecoin demand creates broader ecosystem token demand independent of the emission mechanics. The combination is designed to produce sustainable token economic dynamics even as the emissions continue.

The empirical evidence about whether this works will only be available over a longer time horizon than the protocol has yet operated. The first year of mainnet has supported substantial activity, but the structural sustainability question is whether the model continues to produce attractive economics for participants after the initial enthusiasm and emissions-driven activity matures.

The Comparison to Other Liquidity-First L1 Approaches

Berachain’s Proof-of-Liquidity approach can be compared to other Layer 1 attempts to address the cold-start liquidity problem through specific tokenomic mechanisms. The ve(3,3) approach that Aerodrome and similar DEXes have used shares some conceptual similarities to Proof-of-Liquidity in directing emissions through a vote-escrow mechanism that creates structural participation incentives.

The differences are important. Aerodrome operates as a DEX application within a broader Layer 2 ecosystem (Base), while Berachain attempts to apply similar incentive concepts at the Layer 1 consensus level. The integration of liquidity provision with consensus security is a more ambitious architectural choice than applying liquidity incentive mechanisms to a single application. The success or failure of Berachain’s specific approach therefore tests a different hypothesis than the success of vote-escrow DEX approaches has tested.

Other Layer 1 approaches that have prioritised liquidity bootstrapping include the various incentive programmes that Solana, Avalanche, and other major chains have run at different points to attract DeFi activity. These have generally been time-limited incentive programmes rather than structural protocol features, which means the activity they generated was often temporary rather than sustained. Berachain’s bet is that structural integration of liquidity incentives with consensus security produces more durable activity than time-limited incentive programmes.

The Competitive Positioning

The competitive landscape that Berachain operates within includes the established Layer 1s (Ethereum, Solana), the leading Ethereum L2s (Arbitrum, Base, Optimism), and the other newer Layer 1 challengers (Sui, Aptos, Monad). The specific niche that Berachain has positioned for — being the DeFi-first Layer 1 with strong liquidity incentive mechanisms — overlaps with several of these competitors in different ways.

Against Ethereum and the Ethereum L2 ecosystem, Berachain competes for the DeFi developer attention and for the liquidity that DeFi applications require. The Ethereum ecosystem has substantially more developer talent, more mature applications, and more established institutional integration than Berachain has been able to build in its first year of operation. The Berachain proposition is that the specific liquidity incentive mechanisms produce competitive advantages that the Ethereum ecosystem cannot match.

Against Solana, Berachain faces a competitor that has substantial DeFi activity, strong developer ecosystem, and the post-ETF institutional credibility that Solana has built. Solana’s established DEX volume and DeFi ecosystem represent direct competitive overlap with the categories that Berachain has positioned for.

Against the other Layer 1 challengers, Berachain has competed reasonably for the share of DeFi-focused activity that is open to newer Layer 1 options. The relative success across the Layer 1 challenger cohort has been variable, with different protocols winning in different specific niches. Berachain’s specific position in the DeFi-first category has been one of the more visible niches that newer Layer 1s have established.

The Honest Assessment for Investors and Participants

For investors evaluating Berachain exposure (BERA token, BGT token, or specific ecosystem application exposure): the protocol represents a genuine innovation in Layer 1 tokenomic design, the early ecosystem development has been substantial, and the structural sustainability questions remain open in ways that affect the appropriate risk sizing of any specific exposure.

The bull case for Berachain rests on the Proof-of-Liquidity mechanism producing sustained ecosystem development that other protocols cannot replicate, the BGT emissions creating ongoing demand from protocols seeking emissions that supports the token economics, and the broader ecosystem developing the kind of organic activity that justifies the structural design choices. The bear case is that the emission-driven activity that has supported the early ecosystem development is not sustainable as emissions normalise, that the complex three-token structure produces operational friction that limits ecosystem growth, and that the broader Layer 1 competition leaves Berachain in a niche that cannot scale to the level that the current valuations imply.

The probable outcome is somewhere between these scenarios. The protocol has produced enough innovation and ecosystem development to establish a meaningful position in the broader Layer 1 landscape, but the eventual scale of that position depends on how the tokenomic sustainability questions resolve over the next several years. The next 12-24 months will provide important empirical evidence about whether the Proof-of-Liquidity model produces sustained activity at scale or whether the initial enthusiasm proves difficult to sustain.

For DeFi participants evaluating ecosystem participation on Berachain: the bribe market dynamics provide opportunities for yield generation that may not be available on other chains, the specific incentive mechanisms can be lucrative for participants who understand the system, and the broader ecosystem development provides opportunities for early positioning in applications that may grow over time. The risks include the structural questions about the underlying tokenomic sustainability and the specific risks of participating in DeFi protocols that depend on continued BGT emissions for their economic attractiveness.

The honest position is that Berachain represents one of the more interesting Layer 1 experiments of the current cycle, that the initial results have validated the basic feasibility of the Proof-of-Liquidity approach, and that the long-term sustainability is still being tested in ways that require continued observation. The protocol has earned the attention that it has received through genuine innovation; whether that innovation translates into sustained competitive position will be determined by execution and by the broader market dynamics that affect all Layer 1 protocols.

The Product Question Underneath the Protocol: What Berachain’s PoL Is Actually Asking Users to Do

Julie Zhuo’s framework for evaluating products starts with a simple question: what is the user actually being asked to do, and is that ask proportionate to the value they receive in return? Applied to Berachain’s Proof-of-Liquidity mechanism, the question produces a useful clarification. PoL is not just a consensus mechanism. It is a user experience design choice that determines who participates, why they participate, and whether the participation produces the network effects that the design depends on.

What PoL asks users to do is more involved than standard staking. A validator on Berachain does not simply lock tokens and earn yield. The validator directs block rewards to liquidity pools of their choosing, and those pools earn BGT — the non-transferable governance token that is the actual scarce resource in the system. Users who want BGT must provide liquidity to the pools that validators favour. Validators who want delegation must earn the trust of the BGT holders who will boost their weight. The mechanism creates a multi-step engagement loop that is significantly more complex than depositing into a yield vault.

The Maker Sky Endgame transformation provides one reference point for what happens when DeFi protocol design requires multi-step user engagement. Maker’s governance system — DAI stability fees, collateral onboarding votes, Endgame restructuring — has historically suffered from low participation relative to total token supply. The users who understand the mechanism well enough to participate actively are a small fraction of those holding the token. PoL’s engagement requirement is more integral to the protocol than Maker governance, but the participation ceiling imposed by complexity is a real constraint on how widely the mechanism can distribute rewards.

MEV dynamics interact with PoL in ways that have not yet been fully stress-tested at scale. The validator’s ability to direct block rewards creates an information advantage about which liquidity pools will receive BGT emissions. A sophisticated validator — or a block builder with an information relationship with validators — can position in those pools before the emissions are announced and extract the price impact of the incoming liquidity. This is a form of MEV that is native to PoL’s design rather than being an artefact of the execution environment. How the Berachain team addresses this will be a significant determinant of whether large-scale liquidity provision by sophisticated participants is net-positive or net-extractive for retail participants.

stablecoin B2B payment infrastructure is central to Berachain’s liquidity bootstrapping. The pools that validators direct rewards to are predominantly stablecoin pairs and BERA/stablecoin pairs. The depth of stablecoin B2B infrastructure — the rails that allow institutional participants to move large USDC or USDT positions into DeFi efficiently — directly determines how quickly Berachain’s liquidity pools can reach the depth required for meaningful trading volume. A protocol whose liquidity mechanism depends on stablecoin depth is implicitly dependent on the maturity of stablecoin infrastructure more broadly.

the crypto privacy renaissance is relevant to Berachain’s long-term positioning in a specific way. If ZK-enabled privacy for DeFi transactions becomes available on Berachain or as a composable layer above it, the validator incentive structure changes. Private liquidity provision — where the validator cannot observe which pools are attracting sophisticated capital — reduces the information advantage that the current PoL design creates. Whether that is a feature or a bug depends on your view of what the mechanism is optimising for.

ECB-Fed policy divergence matters for Berachain’s growth trajectory in the same way it matters for all new DeFi protocols: global risk appetite determines how much speculative capital is available for new mechanism experiments. A rate environment that pushes capital toward yield generates interest in PoL’s emissions structure. A risk-off environment reduces the marginal buyer for BERA and BGT regardless of the protocol’s technical merits.

Berachain has designed something genuinely novel. Whether it is genuinely useful at scale depends on whether the engagement requirement can be made proportionate to a wide enough user base to generate the network effects the design requires.

The PM’s Read on Proof-of-Liquidity: What the Protocol Is Asking Users to Do and Whether They Will Do It

Julie Zhuo’s product management framework begins with the user’s perspective rather than the builder’s perspective: what is the user being asked to do, what problem does that action solve for them, and what would make them more likely to do it consistently? Applied to Berachain’s proof-of-liquidity mechanism, the product management question is not whether the mechanism is technically elegant or economically novel — it is whether the validator, the liquidity provider, and the end user are each being asked to do something that aligns with their existing motivations rather than something they have to be incentivized away from their natural behavior to do.

The validator’s job-to-be-done in the proof-of-liquidity system is to decide where to direct BGT emissions across the incentivized liquidity pools. The PM’s lens asks: is this a decision that validators are equipped to make well, and what happens when they make it poorly? The validator is being asked to evaluate the productive value of competing liquidity pools and direct emissions accordingly — a task that requires the same analytical capability as a VC making capital allocation decisions, applied to on-chain liquidity pools rather than companies. The validator who makes this decision well (directing emissions to pools where the liquidity actually creates network value) produces better outcomes for the ecosystem than the validator who makes it poorly (directing emissions to pools where they have financial relationships that may not align with ecosystem productivity). The system’s health depends on whether the validator incentive to capture BGT value aligns with the validator’s incentive to direct emissions productively — which is the product design question that distinguishes PoL from simpler validator reward mechanisms.

The liquidity provider’s job-to-be-done is to deposit assets into the pools where the BGT reward is sufficient to justify the impermanent loss and counterparty risk exposure. The PM’s lens asks: is the information available to the LP sufficient to make this decision well? The LP needs to understand not just the current BGT emission rate to a pool but the expected future emission rate (which depends on validator decisions that are not predictable with certainty), the impermanent loss risk given the pool’s asset composition and historical volatility, and the borrow/lending risk if the pool is connected to lending infrastructure. This is a more complex decision than a simple yield optimization, and the LP’s ability to make it well depends on the quality of the interface and analytics that the ecosystem provides. Enterprise AI adoption faces the same PM challenge: the feature set is sophisticated and the potential value is real, but the interface complexity for the non-technical enterprise user makes the gap between “could use” and “does use regularly” very wide. The 3.3% penetration is the LP-equivalent problem at the enterprise software layer — the decision to engage is complex enough that most users who could benefit do not.

The end user’s job-to-be-done — the DApp user who interacts with the Berachain ecosystem through the liquidity that PoL enables — is the simplest test of whether the mechanism produces real-world value. The end user should experience better liquidity, lower slippage, and more reliable execution than they would on an alternative chain, as a direct result of the BGT incentive mechanism directing capital productively into the pools they use. If the end user experience is not better than the alternative, the mechanism’s theoretical elegance is irrelevant — the product has failed the basic PM test. VC investment in Berachain ecosystem applications is the signal that the VC layer believes the end user experience test will be passed — but VC belief is a stated preference, not a behavioral signal. Independent evaluation of protocol user experience — editorial coverage that assesses what the protocol actually delivers rather than what it promises — is the Zhuo-standard behavioral evidence that distinguishes genuine product-market fit from VC-funded promotional adoption. Friction in the PoL participation path is the mechanism that will determine whether Berachain’s theoretical alignment between validator incentives and ecosystem productivity translates into actual behavioral alignment: every step in the process where the required action is complex, opaque, or requires active management is a friction point where the intended behavior diverges from the actual behavior. Prediction markets on Berachain’s active daily addresses at six months post-mainnet are pricing a wider range of outcomes than the VC-backed promotional narrative implies — which is the PM framework’s honest acknowledgment that the gap between mechanism elegance and user adoption is determined by the friction reduction work that happens after launch, not by the mechanism design that preceded it.

The Moat Architecture

The AI coding assistant category is unusual in that adoption is running years ahead of differentiation. Every significant enterprise AI deployment study shows developers as the cohort most willing to pay for AI tools — and yet no single vendor has established what Seven Powers analysis would call a durable competitive moat. Cursor has strong product velocity and an engaged developer base, but switching costs remain low because the underlying foundation models are available to competitors at commodity pricing. GitHub Copilot has the distribution advantage of the Microsoft-GitHub-Azure stack, but its core product has consistently lagged on user preference rankings among professional developers. Windsurf demonstrated that a well-executed new entrant could take meaningful market share in under twelve months, which is the clearest possible signal that this category has not yet stabilised around a structural winner. The most likely path to durable Power is institutional data: enterprises that build proprietary codebases and internal knowledge bases on top of a particular assistant accumulate switching costs over time as the tool learns their conventions and architecture. The shift to GitHub Copilot usage-based billing 2026 is the clearest signal that Microsoft sees this dynamic — unit economics tied to agentic task completion rather than seat count align revenue with the value creation that generates lock-in. Whether Copilot can rebuild its product reputation before competitors replicate the data-flywheel strategy is the defining competitive race in this category over the next eighteen months.

AI coding assistants have emerged as the most successful enterprise AI deployment category by the most meaningful metrics: actual production usage by paying customers, sustained revenue growth, and the share of engineering teams that have integrated AI coding tools into their daily workflows. Where most enterprise AI use cases remain stuck in pilot evaluations or limited production deployments, AI coding assistants are operating at substantial scale across software engineering teams ranging from startup-stage to the largest enterprises.

The competitive landscape has crystallised around several distinct categories. GitHub Copilot, owned by Microsoft and powered by a combination of OpenAI and proprietary models, remains the largest deployed AI coding assistant by user count and continues to integrate deeply with the broader Microsoft developer ecosystem. Cursor — the IDE-first AI coding assistant — has grown rapidly to over half a billion in annualised revenue and represents the strongest case study for AI-native developer tools. Windsurf (formerly Codeium) has positioned itself as the enterprise-focused alternative with stronger compliance and on-premises deployment options. Devin from Cognition AI represents the autonomous agent end of the spectrum — AI that operates more independently to complete coding tasks. Several other entrants — Continue, Tabnine, Replit’s AI products, Anthropic’s own Claude Code — round out a competitive market that has more credible players than any other enterprise AI category.

Understanding what the AI coding assistant category actually reveals about enterprise AI adoption requires looking at the specific competitive dynamics, the value chain economics, and the structural questions about which categories of AI tools generate the most durable customer relationships.

Why AI Coding Adoption Worked Where Other Enterprise AI Has Stalled

The AI coding assistant adoption pattern stands out compared to other enterprise AI categories that have struggled to convert from pilots to production. Several structural factors explain why coding has been the breakthrough use case.

The output of an AI coding assistant — code that the developer can immediately review, test, and incorporate — has an easy evaluation mechanism. A developer can quickly assess whether a code suggestion is helpful, partially helpful, or wrong, and the cumulative experience of these evaluations produces clear feedback about whether the tool is providing value. This is different from many other enterprise AI use cases (customer support automation, document analysis, business intelligence summarisation) where evaluating the quality of AI output is harder and slower.

The deployment friction for AI coding tools is also significantly lower than for other enterprise AI categories. A developer can install a coding assistant as an IDE extension or sign up for a SaaS tool with minimal IT involvement, evaluate it personally, and make individual adoption decisions. Enterprise procurement and IT review eventually catches up for compliance and security purposes, but the initial adoption typically happens through individual developer choice rather than top-down IT decisions. This bottom-up adoption pattern accelerates the proof-of-value cycle considerably.

The productivity gains from AI coding assistants are also clearly attributable to the tool in ways that other enterprise AI productivity claims are not. A developer using an AI coding assistant who reports completing 30 percent more pull requests can connect that productivity to the tool through specific examples — code that was generated, refactored, or debugged with AI assistance. The same productivity claims for AI-augmented sales operations or marketing functions are harder to measure and harder to attribute.

The broader AI safety considerations for code generation have also matured significantly as the category has scaled. Concerns about AI-generated code introducing security vulnerabilities, license violations, or inferior architectural decisions have been addressed through deployment patterns that emphasise developer review of AI suggestions, code scanning integration, and the broader software development lifecycle controls that organisations already maintain.

Cursor and the IDE-First Strategy

Cursor has been the most discussed case study in the AI coding assistant category, growing from a 2023 launch to over half a billion in annualised revenue by 2026 — one of the fastest revenue ramps in the SaaS industry’s history. The product’s positioning is straightforward: a complete IDE built around AI assistance rather than an AI assistant grafted onto an existing IDE. The user experience differences from Copilot-in-VS-Code are subtle but significant for developers who heavily use the AI capabilities — the interactions are smoother, the context awareness is broader, and the tool feels designed for AI-augmented workflows rather than adapted to them.

The strategic question for Cursor is whether the IDE-first positioning is sustainable as Microsoft continues to improve GitHub Copilot’s integration with VS Code (which Microsoft owns) and as the underlying model capabilities continue to converge. Cursor’s competitive advantage rests partly on product execution velocity (continuous improvements at a pace that Microsoft’s larger organisation finds harder to match) and partly on the IDE itself becoming a differentiated product that developers prefer for non-AI reasons.

The revenue growth trajectory and the user retention metrics that have been disclosed by Cursor suggest that the customer relationship is durable at least over the timescales relevant for venture investment decisions. Whether the IDE-first strategy produces the multi-decade developer platform position that Microsoft has built with Visual Studio and VS Code is a question that will be answered over much longer timescales.

GitHub Copilot and the Microsoft Platform Advantage

GitHub Copilot continues to operate with the structural advantages that Microsoft’s platform position provides. The integration with VS Code (the most-used developer environment), with GitHub (the dominant code hosting platform), and with the broader Microsoft 365 enterprise relationships gives Copilot distribution that pure-play AI coding assistants cannot easily replicate. The enterprise procurement process for Microsoft products often includes Copilot as part of broader software agreements that simplify the adoption decision for IT organisations.

The criticism of GitHub Copilot from developers has been that the product has been less aggressive in adopting cutting-edge AI capabilities than the dedicated AI-first competitors. The pace of Copilot’s feature releases has been slower than Cursor’s, the model integrations have been less timely with the latest model capabilities, and the user experience has been described as feature-conservative compared to the AI-native alternatives. Microsoft’s strategic response has been to accelerate Copilot’s development through deeper integration with internal AI capabilities and through specific feature investments (Copilot Workspace for project-level AI capabilities, deeper agent integrations) that aim to close the perceived gap.

The competitive dynamic between Microsoft Copilot and the AI-first alternatives mirrors many prior cycles in enterprise software, where the incumbent platform leverages distribution to maintain market share while pure-play challengers innovate on product. The historical pattern is that distribution generally wins for the broader market while pure-play challengers capture the segments that most value product innovation, which is consistent with what is happening across the AI coding assistant category.

Devin and the Autonomous Agent Frontier

Devin, developed by Cognition AI, represents a different category from the AI coding assistants discussed above: rather than augmenting a developer’s individual coding work, Devin operates as an autonomous coding agent that can be given high-level task descriptions and that completes those tasks across multiple files, potentially across multiple sessions, with limited human intervention. The product positioning is that Devin operates more like a junior engineer who can be assigned tickets than like an autocomplete tool that assists a senior engineer.

The honest assessment of autonomous coding agents in 2026 is that the capability is genuinely impressive in specific scenarios but unreliable enough that production deployment requires careful task selection and review. Tasks that are well-scoped, that have clear acceptance criteria, and that operate within familiar codebases can be completed by Devin with reasonable success rates. Tasks that are ambiguously specified, that require significant architectural decisions, or that involve unfamiliar codebases produce results that often require substantial human rework.

The competitive dynamic at the autonomous agent end of the spectrum includes Devin, Claude Code’s autonomous capabilities, GitHub Copilot’s evolving agent features, and several other entrants. The category is rapidly evolving and the specific products that achieve sustained market positions will likely be determined by both capability improvements and by which providers solve the operational challenges of running autonomous coding work reliably at enterprise scale.

The Value Chain and Where Margins Actually Sit

The AI coding assistant value chain provides a useful case study in where AI-era enterprise software value actually accrues. The chain includes the underlying foundation model providers (OpenAI, Anthropic, Google, Meta), the AI coding assistant products that integrate those models into developer-facing tools (Cursor, Windsurf, Copilot, Devin), and the infrastructure providers that enable the deployment (cloud providers, GPU infrastructure, training compute).

The model providers capture significant value through API revenue from the coding assistant products that integrate their models. OpenAI’s API revenue has been substantially supported by coding-related usage, and Anthropic has positioned Claude as particularly strong for coding use cases with corresponding API revenue benefits. The dynamic is that the coding assistant products must pay the model providers for the underlying API calls, which compresses the gross margins of the coding assistant products themselves.

The coding assistant products at the application layer have varied unit economics depending on their pricing model, customer mix, and operational efficiency. Cursor’s reported revenue at high gross margins suggests that the application layer can be profitable when pricing power supports the margin requirements, but the structural pressure from foundation model costs is real and persistent.

The infrastructure layer — cloud providers running the AI workloads, GPU infrastructure supporting model training and inference — captures the largest absolute value in the chain because the compute requirements for coding-related AI workloads are substantial and growing. Nvidia’s continued dominance in AI compute means that the infrastructure layer revenue concentrates in a small number of beneficiaries who capture the demand that the application layer creates.

What This Reveals About Enterprise AI More Broadly

The AI coding assistant success provides useful evidence about which enterprise AI use cases are likely to scale and which face structural challenges. The categories with similar characteristics — easy output evaluation, low deployment friction, clear productivity attribution, bottom-up adoption potential — are more likely to follow the same successful trajectory. The categories without these characteristics — complex output evaluation requiring extensive human review, top-down deployment requirements with significant IT coordination, productivity claims that are difficult to attribute to the AI specifically — face structural adoption challenges that the coding assistant pattern does not provide a roadmap for.

The agentic AI threats to enterprise SaaS need to be evaluated against this framework. AI agents that automate specific, evaluable tasks within established workflows are more likely to succeed at scale than agents that aim to replace broader human roles with less clearly defined success criteria. The categories where seat-based SaaS faces real disruption are those where the automated tasks have characteristics similar to what made coding assistants successful.

For investors evaluating enterprise AI exposure: the AI coding assistant category provides the most concrete evidence that enterprise AI can produce substantial revenue businesses with durable customer relationships. The specific companies in the category face competitive dynamics that will determine which capture sustainable positions, but the category itself has demonstrated commercial viability at scale. The transferability of these lessons to other enterprise AI categories is real but conditional on whether those categories share the structural characteristics that enabled coding assistant success.

The honest position is that AI coding assistants are the closest thing the enterprise AI category has to a proven product-market fit, that the lessons from this success help identify which other AI use cases are likely to follow similar trajectories, and that the specific competitive dynamics in the coding assistant market will continue to evolve as the underlying AI capabilities improve and as the platform leaders adapt their strategies to defend their positions.

Following the Money: Where the AI Coding Assistant Revenue Actually Goes

Carl Bernstein’s method is to follow the money past where the press release stops. The AI coding assistant market in 2026 has produced impressive adoption headlines and equally impressive revenue claims. Following the money reveals a value chain where the distribution of that revenue is significantly less favourable to the pure-play assistants than the headline numbers suggest.

GitHub Copilot’s revenue flows to Microsoft. Not to GitHub as an independent entity — GitHub was acquired in 2018 — and not to the model providers whose outputs power the suggestions. Microsoft’s platform control over the developer environment means that Copilot’s adoption is simultaneously growing the revenue line for a company that also controls the IDE, the source control system, the CI/CD infrastructure, and the cloud environment where the code ultimately runs. The value of Copilot to Microsoft is not primarily the $19/month subscription. It is the lock-in of the developer workflow to the Microsoft stack at a moment when developer tooling decisions become ten-year infrastructure choices.

Cursor and Windsurf are collecting subscription revenue directly. The question following the money asks is where the margin sits. Both companies pay inference costs to the model providers — Anthropic, OpenAI, Google — that are not trivial relative to the subscription price at current usage rates. The gross margin on an AI coding assistant subscription is a function of the ratio between inference costs and monthly fee, and at heavy usage, that ratio is uncomfortable. The companies that built the best developer product may not be the companies that built the most durable business, because the model providers sit above them in the value chain and can reprice at will or launch competing products.

enterprise SaaS agentic AI threat is the larger competitive threat that the pure-play coding assistant narrative tends to underemphasise. Salesforce Agentforce, ServiceNow AI, and Workday’s AI layer are all attempting to make the enterprise software suite itself agentic — capable of taking actions, not just suggesting them. If the enterprise software environment becomes agentic, the demand for a separate AI coding layer changes. Developers using Salesforce infrastructure will use Salesforce’s AI tools because the context — the data, the workflow, the permission model — is native to the platform. The standalone coding assistant competes with this only if it has richer context than the native environment. At scale, native environments win on context.

cybersecurity vendor consolidation is an underreported cost centre for AI coding assistant deployments. Enterprise security teams reviewing Copilot or Cursor for deployment must assess whether the code suggestions are leaking proprietary patterns, whether the telemetry sent to the model provider is within data governance requirements, and whether the model’s training data creates IP liability for generated code. These are not hypothetical concerns — they have delayed or prevented enterprise rollout at multiple large organisations. The adoption numbers for AI coding tools in regulated industries are systematically lower than the headline enterprise adoption figures suggest.

Snowflake vs Databricks AI workload competition illustrates the data infrastructure dependency that AI coding tools are now surfacing. Developers working with large-scale data pipelines — Snowflake queries, Databricks notebooks, dbt transformations — need AI assistance that has context about the specific schema, the specific data quality issues, and the specific performance constraints of their environment. Generic code suggestions are less useful than context-aware suggestions. The companies building data-aware coding intelligence are building into a more defensible position than the companies building generic coding assistance, because the context advantage compounds with usage.

Q2 2026 earnings season preview will provide the first systematic evidence of whether AI coding tools are appearing in corporate cost lines as productivity investments or as experimental discretionary spend. The distinction matters: productivity investment is sticky and grows with headcount; discretionary spend is the first thing cut when margin pressure arrives. Following which line item AI coding tool costs appear in, and whether they appear in capex or opex budgets, tells you more about the category’s durability than the adoption survey data does.

The headline adoption story is real. The revenue durability story requires more scrutiny than the headlines provide.

Crypto privacy technology has experienced a longer and more difficult period than almost any other crypto category over the past several years. The combination of regulatory pressure on privacy-enabling tools, exchange delistings of privacy coins, banking restrictions on transactions involving privacy protocols, and the technical challenges of making privacy-preserving systems usable for ordinary applications produced a multi-year period during which privacy seemed to be a category in retreat rather than advancement.

The picture in 2026 is meaningfully different. Zero-knowledge proof technology has matured into production-ready infrastructure that supports a generation of privacy products that are genuinely more usable than their predecessors. Zcash’s continued protocol development has produced significant performance improvements. Aleo has launched mainnet and built an early application ecosystem. Aztec’s privacy-preserving Layer 2 on Ethereum has attracted developer attention. ZK rollups for general computation have made privacy a deployable feature rather than a research aspiration. The combination of technical maturity and institutional recognition that financial privacy is a legitimate requirement — not just for criminal use cases — has produced what the privacy community calls a renaissance.

Understanding what is actually working in crypto privacy in 2026 requires distinguishing the technologies that have matured into deployable products from the broader privacy narrative, and recognising both the regulatory constraints and the genuine institutional interest that are shaping the category’s trajectory.

What Zero-Knowledge Proofs Actually Enable

The underlying technology that powers most of the 2026 privacy renaissance is the dramatic improvement in zero-knowledge proof systems over the past five years. Zero-knowledge proofs allow a party to prove that a statement is true without revealing the underlying information that makes it true — for example, proving that a transaction is valid and that the sender has sufficient balance without revealing the sender’s address, the recipient’s address, or the transaction amount.

The technical advances that have made this practical at scale include the dramatic reduction in proof generation times (from minutes for early zk-SNARK systems to seconds or sub-second for current implementations), the development of zkVMs (zero-knowledge virtual machines) that allow general-purpose computation to be proved rather than only specific predefined operations, and the maturation of hardware acceleration for proof generation that makes the systems competitive on cost compared to non-private alternatives.

These improvements have moved zero-knowledge proofs from a research curiosity to a deployable component in production systems. The same technology that underpins Ethereum’s scaling roadmap through ZK rollups enables privacy applications that use proof verification to maintain confidentiality while still providing public verifiability of correctness. The infrastructure investment that the broader crypto industry has made in ZK technology has produced benefits for privacy applications that would have been impractical without it.

Zcash and the Older Privacy Coin Story

Zcash represents the most established privacy coin and has continued to develop its protocol despite the broader headwinds the category faced. The Halo 2 implementation eliminated the trusted setup ceremony that earlier zk-SNARK constructions required, the protocol has continued to add capabilities to its shielded transaction infrastructure, and the user experience has improved through wallet developments like Zashi and integrations with mobile-friendly Zcash applications.

The honest assessment of Zcash’s market position is mixed. The technology has improved meaningfully, and the privacy guarantees that shielded Zcash transactions provide are among the strongest in any production cryptocurrency. The market capitalisation and trading volume reflect the regulatory friction the asset has faced — exchange delistings in multiple jurisdictions limited the addressable market and concentrated remaining trading in the venues that supported the asset. The user base remains committed but smaller than the technology would arguably justify.

The strategic question for Zcash is whether its existing technical advantages and brand recognition can support a renewed adoption phase as the broader privacy renaissance attracts new users to the category. The competition has expanded — Aleo, Aztec, and various ZK rollup-based privacy applications all compete for the user interest that Zcash historically would have captured by default — and Zcash’s response depends on the protocol’s ability to differentiate on technical capability and user experience.

Aleo and Programmable Privacy

Aleo represents the second-generation privacy protocol approach: a Layer 1 blockchain designed specifically for privacy-preserving applications, with native support for confidential computation through its Leo programming language. Where Zcash provides privacy for the specific use case of value transfer, Aleo extends privacy to arbitrary application logic — developers can build applications where the computation itself, the inputs to the computation, and the outputs can all be selectively private while still being verifiable on-chain.

The applications that this enables go beyond simple private payments. Private auctions where bids remain confidential until execution, private voting where individual votes are confidential but the aggregate tally is verifiable, private DeFi positions where holdings and trading activity are not publicly visible, and private identity applications where credentials can be verified without exposing the underlying identity information. The breadth of potential applications is significantly larger than what privacy coins alone enable.

Aleo’s challenge in 2026 is the standard challenge for any new Layer 1: bootstrapping a developer ecosystem and an application layer that demonstrates the privacy capabilities in production use cases. The early Aleo application ecosystem includes several promising developments but is in the same early-stage position that other Layer 1 challengers face when competing for developer attention against established platforms.

Aztec and Privacy on Ethereum

Aztec represents a different approach: rather than building a privacy-focused Layer 1, Aztec is building a Layer 2 on Ethereum that provides privacy as a feature within the broader Ethereum ecosystem. This positioning leverages Ethereum’s existing developer ecosystem, liquidity, and infrastructure while adding privacy capabilities through Aztec’s zero-knowledge proof architecture.

The strategic appeal of the Aztec approach is that it does not require users to migrate their assets and applications to a new Layer 1 — they can use Ethereum-native applications and selectively access privacy through Aztec when specific transactions or applications require it. The integration with the broader Ethereum ecosystem provides Aztec with structural advantages in attracting developers and users that pure-play privacy Layer 1s do not have.

The mainnet launch of Aztec’s privacy Layer 2 in 2025 has been followed by early ecosystem development that demonstrates the architecture works at meaningful scale. The applications that have launched on Aztec include privacy-preserving DeFi protocols, private payment infrastructure for enterprise use cases, and identity applications that integrate with Ethereum-based credential systems.

The Regulatory Environment and Institutional Interest

The regulatory environment for crypto privacy has been historically hostile but is evolving in important ways. The early 2020s saw aggressive regulatory action against privacy tools, including the Treasury’s OFAC sanctions of Tornado Cash and the prosecutions of mixer service operators. These actions chilled the broader privacy infrastructure development and led to exchange delistings of privacy coins in multiple major jurisdictions.

The 2026 regulatory landscape is more nuanced. The recognition that financial privacy is a legitimate requirement for institutional use cases — corporate treasury management, M&A activity, supply chain payments where competitive sensitivities matter — has produced a regulatory conversation that distinguishes between privacy for criminal activity (which remains targeted by enforcement) and privacy for legitimate financial activity (which has begun to be acknowledged as a category that regulatory frameworks need to accommodate).

The technical development of selective disclosure mechanisms has been important in shifting the regulatory conversation. Zero-knowledge proof systems can be designed to provide cryptographic guarantees of compliance with specific requirements — proof that an address is not on a sanctions list, proof that a transaction amount is below regulatory reporting thresholds, proof that the participants in a transaction have completed KYC at an appropriate gateway — without revealing the underlying transaction details. The combination of privacy with provable compliance is a more regulator-friendly framing than the absolute privacy that earlier privacy coins emphasised.

Institutional interest in privacy technology has grown as enterprises have recognised that tokenised real-world assets and institutional DeFi participation require privacy that public blockchain transparency does not naturally provide. A bank that wants to deploy capital into on-chain lending markets cannot have its trading and position information publicly visible to competitors and counterparties. A corporate treasurer managing tokenised cash positions cannot have the company’s liquidity profile visible to all market participants. The institutional use case for privacy is genuinely growing and is a more legitimate driver of privacy technology adoption than the historical retail-focused privacy coin narrative.

The Honest Assessment for Investors and Builders

For investors evaluating exposure to the privacy technology renaissance: the category is real, the technical progress is substantial, and the institutional demand drivers are credible. The specific projects within the category have very different risk-return profiles. Zcash is the established asset with the strongest brand and weakest growth momentum. Aleo is the most direct play on programmable privacy with the typical risks of an early-stage Layer 1. Aztec benefits from Ethereum ecosystem integration but depends on the broader Ethereum L2 dynamic. ZK rollup-based privacy applications represent a more diffuse exposure across the broader ZK ecosystem.

For developers building on privacy infrastructure: the user experience and developer tooling for ZK applications has improved substantially but remains harder than building non-private applications. The selective disclosure infrastructure that enables compliant privacy is genuinely valuable for institutional applications but adds engineering complexity that consumer applications often do not justify.

For end users — both individuals seeking financial privacy and institutions seeking confidential transaction infrastructure — the 2026 environment offers genuine improvements over what was available three years ago. The privacy guarantees are stronger, the user experience is better, and the regulatory acceptance of privacy as a legitimate category is gradually improving. The renaissance is real, even if it is being driven by less politically dramatic forces than the early privacy coin advocacy implied.

Is ZK Privacy Structurally Better, or Differently Flawed?

The mental model most people apply to privacy technology is a spectrum: more privacy is better, less privacy is worse, and the question is how far along the spectrum a given technology sits. That framing explains why ZK proof-based privacy is so easy to sell as an advance — it is measurably more private than the alternatives that preceded it. What it misses is the structural question: does a better privacy mechanism solve the right problem, or does it solve one dimension of the problem while leaving a different failure mode intact?

Consider the pattern across privacy-related regulatory frameworks. The GDPR — widely acknowledged as the most consequential privacy regulation in modern history — established clear consent requirements, data minimisation principles, and user rights that were technically superior to what preceded them. Seven years after enforcement began, the documented evidence on outcomes is mixed: large platforms with compliance infrastructure adapted relatively smoothly, enforcement against smaller actors has been inconsistent, and the fundamental problem the regulation was designed to address — opaque data practices by powerful commercial entities — has persisted in modified forms. The gap between what GDPR requires and what the enforcement architecture actually produces is not a failure of the regulation’s design. It is a structural property of any rule-based privacy framework operating against adversarial economic incentives.

The same mental model applies to ZK privacy technology. Zero-knowledge proofs are genuinely superior to the privacy mechanisms they replace — more mathematically rigorous, more resistant to side-channel analysis, capable of providing verifiable compliance guarantees alongside privacy. These are real improvements. The structural question is whether ZK privacy solves the right failure mode. If the primary failure of previous privacy technology was insufficient cryptographic strength, ZK is the solution. If the primary failure was that privacy mechanisms were technically sound but adoption was weak because usability was poor, ZK makes partial progress — current ZK application UX is better than it was three years ago but remains substantially harder than non-private alternatives. And if the primary failure was that privacy tools were adopted by high-risk users but rejected or excluded by the regulated institutions that control the largest capital pools, then ZK’s advance on cryptographic sophistication does not address the adoption barrier.

The most useful frame for investors and builders evaluating privacy technology is not “how strong is the privacy?” but “which failure mode is this addressing, and is that the binding constraint?” For consumer applications in developed markets, the binding constraint is usually usability, not cryptographic strength — users tolerate surveillance from familiar platforms at a level they would not accept from strangers. For institutional applications, the binding constraint is regulatory acceptability, which ZK’s selective disclosure mechanisms directly address. For emerging market applications, the binding constraint may be network effect and liquidity, which neither ZK nor any other cryptographic advance can resolve. Separating these questions produces a more calibrated picture of where the privacy renaissance will actually deliver value versus where it will produce sophisticated technology with limited adoption.

The Fragility Test: Where ZK Privacy Technology Breaks Under Stress

There is a way to evaluate privacy technology that the optimistic literature consistently avoids. The question is not whether ZK proofs work under normal conditions. They do. The question is what happens under adversarial conditions — regulatory pressure, exchange delistings, network-level surveillance — and whether the technology is fragile, robust, or genuinely antifragile in those conditions.

Zcash provides the clearest historical test. Shielded transactions work cryptographically. The privacy guarantee is real in the mathematical sense. What the Zcash experience reveals under stress is that the privacy technology is only as durable as the distribution layer. When major exchanges delisted ZEC — Coinbase in 2023, Binance in 2024 — the shielded pool’s privacy benefits became inaccessible to the majority of users who had entered via centralised exchange rails. The technology survived the delisting intact. The user’s practical access to the technology did not.

This is a fragility pattern with a specific structure: the cryptographic layer is robust, but the sociotechnical system — the combination of the protocol, the distribution infrastructure, and the regulatory environment — is fragile in the way that complex systems are always fragile. The failure point is not the strongest component. It is the component with the highest sensitivity to external shocks.

Binance’s effective suspension from European markets under MiCA’s fit-and-proper test illustrates how regulatory stress cascades through the exchange distribution layer to affect the entire DeFi and privacy technology stack. When the dominant exchange infrastructure in a jurisdiction becomes unavailable, every asset that depends on that distribution rail — including privacy tokens — loses access to its primary liquidity channel. The ZK proof that makes a transaction private does not help a user who cannot find a liquid market in their jurisdiction.

The antifragility test for ZK privacy technology asks a different question: does the technology gain from this stress? The honest answer, at this stage of development, is partially. Regulatory pressure on centralised exchange distribution has accelerated development of decentralised exchange infrastructure for privacy assets. Aztec’s privacy-preserving DeFi applications and Aleo’s programmable privacy model gain relevance precisely because the centralised distribution model is under pressure. The stress is generating architectural evolution.

What the investor should track is not the current performance of ZK privacy systems under normal conditions, but the stress test results: how did adoption survive the 2023-2024 delistings? What did the user base look like after the distribution layer shocks? Which protocols retained developer commitment through regulatory uncertainty? The answers to those questions reveal the genuine fragility/robustness profile of each project more accurately than the technical capability claims in any whitepaper. Technology that survives adversarial conditions intact — and ideally grows stronger through them — is worth examining. Technology that is theoretically superior but operationally dependent on a compliant distribution infrastructure is a conditional bet on regulatory tolerance that has not been consistently offered.

The tokenized real-world asset market crossed twenty billion dollars in total value in 2026, making it one of the fastest-growing segments in both traditional finance and crypto simultaneously. BlackRock’s BUIDL fund — a tokenized money market fund deployed on Ethereum — surpassed five billion dollars in assets. Ondo Finance’s OUSG product and Franklin Templeton’s BENJI fund demonstrated that regulated asset managers can distribute tokenized short-duration instruments with operational credibility. The proof of concept phase is over. The harder question is whether the market can scale from twenty billion to two hundred billion, and what needs to be true for that to happen.

The honest answer is that the current market success is concentrated in the easiest part of the RWA problem — short-duration government securities that are themselves highly liquid, easy to custody, and simple to price. The hard part of tokenization — private credit, infrastructure debt, real estate, and other genuinely illiquid assets — remains largely unproven at institutional scale, and the gap between the marketing narrative and the operational reality is wider than most coverage of the space acknowledges.

What Is Actually Working: Tokenized Treasuries and Money Market Funds

The demonstrated success case for RWA tokenization is straightforward: take a liquid, short-duration government security or money market fund, wrap it in a blockchain-native token, and make that token accessible to on-chain participants who want yield-bearing dollar collateral. BlackRock’s BUIDL, Ondo’s OUSG, Superstate’s USTB, and similar products solve a real problem in the DeFi ecosystem — the demand for yield-generating collateral that is safer and more stable than algorithmic stablecoins or ETH.

The use case that has driven adoption is DeFi collateral substitution. Protocols like Aave, Morpho, and several institutional DeFi platforms have integrated tokenized Treasuries as eligible collateral, allowing users to earn Treasury yield on their collateral while maintaining borrowing capacity. This is a genuinely new financial primitive: collateral that earns yield passively without the protocol user having to actively manage the underlying investment. For institutional DeFi participants — asset managers, hedge funds, and proprietary trading desks operating on-chain — this is a meaningful operational improvement over holding USDC or USDT as idle collateral.

The relationship between tokenized Treasuries and stablecoins is convergent rather than competitive. A fully reserved, yield-bearing stablecoin that passes interest to holders is functionally similar to a tokenized money market fund. As stablecoin regulatory frameworks like the GENIUS Act require full reserve transparency, the distinction between a compliant stablecoin and a tokenized T-bill narrows. The regulatory and commercial pressure is toward more yield-bearing, more transparent, more auditable forms of on-chain dollar exposure — which is exactly what the current tokenized Treasury products offer.

The Three Problems That Have Not Been Solved

Legal enforceability is the first unsolved problem. A token that represents a claim on a Treasury or money market fund is only as good as the legal structure that makes that claim enforceable across jurisdictions. The leading tokenized Treasury products have robust legal wrappers — BUIDL operates through a regulated investment fund structure; Ondo’s products are issued through regulated entities with established investor protections. But the broader RWA space includes many products where the legal claim is less clear: offshore tokenization platforms, SPV structures in jurisdictions with uncertain digital asset law, and products that claim to represent assets without having tested that claim through a bankruptcy or dispute resolution process.

Secondary market liquidity is the second problem, and it matters most for assets beyond Treasuries. BUIDL and Ondo tokens have reasonable on-chain liquidity because their underlying assets — T-bills and government money market funds — are themselves highly liquid, and the issuers maintain redemption infrastructure. A tokenized private credit loan or real estate equity stake does not have this property. The underlying asset is illiquid; putting it on a blockchain does not create liquidity that did not previously exist. An investor who buys a token representing a stake in a private credit fund and then wants to exit before the fund term ends faces the same liquidity problem they would have with a conventional private credit fund — the blockchain adds settlement efficiency but not secondary market depth.

The private credit market’s existing liquidity challenges are directly relevant here. The secondary market for private credit fund stakes already trades at significant discounts to NAV in stressed environments. Tokenizing those stakes onto a blockchain creates the illusion of improved liquidity through 24/7 trading infrastructure while the fundamental illiquidity of the underlying asset remains unchanged. Regulators and institutional investors who encounter this mismatch in a market stress event will draw the appropriate conclusions about the limits of blockchain-as-liquidity-enhancement.

Interoperability is the third problem. The tokenized RWA ecosystem is fragmented across chains, legal jurisdictions, token standards, and KYC frameworks. BUIDL operates primarily on Ethereum; Franklin Templeton’s BENJI was initially deployed on Stellar and Polygon; other issuers have chosen Solana, Avalanche, or permissioned chains like Provenance Blockchain. A corporate treasurer who wants to use tokenized Treasuries as collateral across multiple DeFi protocols on multiple chains faces a complex operational picture: they need to hold different tokens on different networks, manage cross-chain bridges that introduce their own custody and smart contract risk, and maintain compliance with KYC requirements that vary by issuer and by chain.

Institutional Interoperability Standards and Who Is Winning That Race

The industry has recognised the interoperability problem and is attempting to solve it through standards bodies and cross-chain infrastructure. The DTCC’s Project Whitney has been exploring tokenized securities interoperability with traditional settlement infrastructure. Swift has conducted cross-chain RWA transfer experiments. ERC-3643 and other identity-linked token standards attempt to embed compliance directly into the token rather than relying on off-chain permissioning.

These are meaningful efforts, but they are early-stage. The settlement finality and legal certainty that traditional institutional investors require for large positions does not yet exist across the fragmented tokenized RWA landscape. Institutions that have adopted tokenized Treasuries have done so in controlled conditions — specific products from specific issuers on specific chains with specific legal structures — rather than in the fully interoperable, cross-chain, cross-jurisdiction environment that the long-term vision implies.

The Ethereum ecosystem’s infrastructure evolution is relevant here: as Ethereum’s L2 ecosystem matures and cross-chain messaging improves, the interoperability of Ethereum-native tokenized assets across the L2 landscape improves with it. Ethereum’s regulatory familiarity, its large institutional validator set, and its developer ecosystem give it an advantage as the primary settlement layer for institutional RWA — but that advantage has not yet translated into the seamless cross-protocol interoperability that would unlock the market’s next scaling phase.

What the People Who Built This Market Actually Say

The official narrative around RWA tokenization is consistent across every product launch, conference keynote, and investor deck produced in the last twenty-four months: institutional-grade infrastructure, seamless on-chain settlement, 24/7 liquidity for assets that previously traded in fragmented OTC markets. The marketing language is uniform to the point of interchangeability. What differs — and what the marketing language rarely surfaces — is what the operational documentation actually discloses once you read past the product overview.

Consider redemption mechanics. Every tokenized money market fund promises liquidity. The actual disclosed mechanics of how that liquidity is delivered under stress are substantially more qualified. Redemption is typically gated through the fund administrator, who operates on business-day cycles. Same-day redemption is available only within specific windows. Large redemptions require advance notice. These constraints exist for legal and operational reasons that are legitimate — they mirror the redemption mechanics of the underlying fund structures. But they mean the product’s on-chain representation of near-instantaneous liquidity is a settlement-layer feature, not a redemption-layer feature. The token moves instantly; the cash does not.

The BUIDL product’s disclosed structure, covered in detail in our analysis of BlackRock’s RWA architecture, shows this gap precisely. The product offers token-level transferability on a permissioned basis, but the regulatory wrapper — a Reg D private placement available only to qualified purchasers — means that secondary-market liquidity is structurally constrained by who can legally hold the token. The “liquidity” that tokenization adds is real within the eligible investor set; it does not create the deep, open-market liquidity that the asset class marketing implies.

Custody concentration is the second documented risk that rarely makes the top-line pitch. The overwhelming majority of institutional RWA assets are custodied through a small number of regulated entities — primarily the custodian arms of the same banks that dominate traditional asset custody. Tokenization does not distribute this custody risk; it creates a new technical layer on top of the same concentrated custody structure. If Coinbase Custody is the custodian of record for the underlying assets backing a tokenized fund, the counterparty risk profile of holding that token is, at its base, the counterparty risk of Coinbase Custody. This is disclosed. It is rarely foregrounded in the product narrative.

Legal wrapping complexity is the third area where reported detail diverges from marketing simplicity. Tokenizing a private credit instrument requires creating a legal structure that recognizes the token as the instrument of ownership. The legal opinion chain for doing this across multiple jurisdictions — where token holders may be located in the US, EU, Singapore, and Dubai simultaneously — involves a multi-layered structure of SPVs, master agreements, and jurisdiction-specific waivers that is materially more complex than “put it on chain.” Practitioners who have built these structures describe the legal overhead as the dominant cost driver in RWA tokenization, eclipsing the technical build cost. None of the product marketing mentions this.

The exit mechanics have never been stress-tested at scale. Every tokenized RWA product operating today has operated during a period of relatively stable underlying asset values and normal market conditions. The processes for orderly redemption when underlying assets are under stress — when the private credit instrument is in default, when the real estate fund needs to gate redemptions, when the T-bill rollover faces settlement failure — exist in the documentation but have not been executed under real pressure. The people who built these products know this. Most are careful not to overstate the stress-tested robustness of infrastructure that is, functionally, less than three years old.

None of this means RWA tokenization is fraudulent or that the infrastructure being built is without value. The practitioners building this market are, on the whole, careful about what they claim. The gap is between what careful practitioners say in detailed conversations and what the product marketing says in public. That gap — between the nuanced, operational, constraint-aware description and the simplified, seamless, institutional-grade pitch — is the gap that the next phase of the market will have to close if the institutional investor base is going to commit at the scale the $200 billion projections require.

The $200 Billion Question

The optimistic case for RWA tokenization — a market worth hundreds of billions within three to five years — rests on three developments happening concurrently: regulatory frameworks that clearly govern tokenized securities across major jurisdictions, interoperability standards that allow tokenized assets to move frictionlessly between chains and into traditional settlement infrastructure, and a demonstrated track record for tokenized illiquid assets that generates institutional confidence in the legal and operational model.

None of these three things are fully in place today. The regulatory frameworks are in formation — the GENIUS Act addresses stablecoins, but broader tokenized securities regulation in the US, EU, and Asia is still being developed. Interoperability standards are proliferating without converging on a dominant protocol. And the track record for tokenized illiquid assets requires time and, inevitably, a market stress event that tests the legal and operational infrastructure under conditions it was designed for but has not yet experienced.

The twenty-billion-dollar market that exists today is real and growing. The two-hundred-billion-dollar market is possible but requires the institutional infrastructure to catch up with the blockchain technology. The assets that will drive that scaling are not tokenized T-bills — they are the genuinely illiquid, hard-to-value, complex-to-legally-wrap assets like private credit, real estate, and infrastructure that carry the yield premium that institutions actually want. Whether tokenization can solve the operational and legal challenges of those asset classes, rather than simply making liquid assets marginally more convenient to hold on-chain, is the question the next few years will answer.

The Startup Lens on Tokenized Assets: Which $20 Billion Is Real and Which Is Waiting for the First Stress Test

Paul Graham’s startup analysis framework begins with the question that distinguishes businesses that have found genuine product-market fit from businesses that have found genuine investor interest: do users use it because they want to, or because they are paid to? In the tokenized real-world asset market, this question is more important than the $20 billion AUM figure because a significant fraction of that $20 billion is in products where the incentive to hold the tokenized version of an asset is not primarily the asset’s yield or risk profile — it is the additional incentive layer (points programs, governance tokens, yield boosts) that makes the tokenized version more attractive than the underlying asset in ways that disappear when the incentive program ends. The $20 billion number is real. The question is how much of it would remain without the incentive architecture that currently supports it.

Graham’s product-market fit test for a startup applies directly to the tokenized RWA question: if you removed the incentives tomorrow, how many users would continue using the product because it is genuinely better for their purpose than the alternative? For tokenized Treasury products, the answer is probably most users — the operational convenience of holding a tokenized Treasury in a DeFi wallet while earning yield is a genuine improvement over the alternative for DeFi-native users, and the improvement does not depend on an incentive program to be real. For tokenized private credit products, the answer is less clear — the institutional investor who is holding tokenized private credit for the on-chain yield premium is earning a premium that is partly the credit risk premium (real and persistent) and partly the on-chain novelty premium (real but not persistent). The credit risk premium will remain when the novelty premium compresses; the question is whether the credit risk premium alone is sufficient to sustain the market size that the novelty premium helped build.

Graham’s do-things-that-don’t-scale principle applies to the current phase of RWA tokenization in a way that most market analyses miss: the protocols and products that are winning the $20 billion AUM competition right now are the ones that have built the institutional relationships, custody solutions, and regulatory approvals that are expensive, slow, and non-replicable at scale. BlackRock’s BUIDL fund’s success is not primarily a technology story — it is an institutional relationship story. BlackRock has the custodial infrastructure, the regulatory approval, the institutional client relationships, and the legal structure that makes the tokenized fund acceptable to institutional investors who cannot hold products without these properties. A startup attempting to replicate BUIDL’s success cannot simply build better tokenization technology — it must build the institutional infrastructure that is the actual product, and that infrastructure is the thing that does not scale quickly. On-chain private credit infrastructure is the specific RWA category where the startup approach — build the technology, let the institutions come to you — is most clearly being tested against the institutional relationship approach. The protocols that have built direct institutional relationships with both lenders and borrowers are outperforming the protocols that have built excellent technology while waiting for institutional demand to find them.

Graham’s advice to founders — make something people want, then figure out how to make money — has a specific implication for the RWA market: the products that people genuinely want are the ones solving a real problem that the existing infrastructure creates. The problem that tokenized Treasuries solve for DeFi-native users is real: the ability to hold a yield-bearing stable asset in the same wallet infrastructure as DeFi positions, without leaving the on-chain environment, is a genuine improvement over the existing alternative of holding USDC at zero yield or bridging to TradFi and back. The problem that tokenized private credit solves for institutional lenders is also real but less universally felt: the operational improvements in settlement, reporting, and collateral management are real for large-scale lenders but are not yet compelling enough to overcome the compliance, custody, and counterparty risk concerns for most institutional credit investors. Enterprise adoption job-to-be-done framing is the diagnostic Graham would apply: the RWA product that is being hired for the job the institutional investor needs to do — not the job the protocol builder wants to provide — is the one with durable adoption. Crypto VC allocation to RWA infrastructure is the market’s stated belief about which products solve real problems — but Graham’s framework notes that VC allocation follows narrative as much as product-market fit evidence, and the RWA narrative has been strong enough to attract capital ahead of the behavioral adoption data that would confirm genuine fit. DeFi-native liquidity infrastructure is the on-chain environment that makes tokenized RWA useful to DeFi users — the RWA protocol that integrates into the BGT emission ecosystem has a distribution advantage that the protocol building only for institutional TradFi users does not have. Prediction markets on RWA tokenization AUM at end-2026 are pricing continued growth from $20 billion toward $30-40 billion — which Graham’s framework reads as the market pricing both the genuine product-market fit in Treasuries and the incentive-supported growth in private credit, without adequately distinguishing between the two.

Solana’s fee market has been a persistent source of user frustration since the network became a high-activity environment in 2021. The original design used a single global fee market: every transaction competed for the same block space regardless of which accounts or programs it touched, which meant that congestion in any one application — an NFT mint, a token launch, a liquidation event — created fee spikes that affected every other transaction on the network simultaneously. A DeFi protocol swap would fail or become expensive during an NFT mint in which it had no direct involvement, simply because both were competing for the same undifferentiated block space.

SIMD-0096, the Solana Improvement Document that implemented local fee markets, changes this architecture fundamentally. Rather than a single global fee rate, transactions now pay fees based on the congestion of the specific accounts and programs they access. A mint that is creating massive demand for writes to a particular program’s accounts drives up fees for transactions touching those accounts; it does not affect the fee rate for a DeFi swap touching unrelated accounts. The fee market becomes local to the resources being contested rather than global to all block space.

The announcement of SIMD-0096 was received positively across the Solana developer community, and with good reason — the global fee market problem was real and well-documented. But the practical implications for protocol developers, traders, and infrastructure operators are more nuanced than the announcement framing suggested. Local fee markets solve one problem while introducing several new ones that the ecosystem needs to navigate.

How Priority Fees Work Under the New Architecture

Under the global fee market, priority fees were simple in concept: pay more than the base fee, get priority in block inclusion. The practical complexity was in setting the right amount — too low and the transaction failed during congestion; too high and fees were wasted. Various priority fee estimation services emerged to help applications estimate the current market rate and set fees accordingly.

Under local fee markets, the priority fee estimation problem becomes more complex because the relevant congestion metric is now specific to the accounts and programs a transaction touches. A transaction that writes to a highly contested token account — one involved in a popular DeFi protocol — needs to estimate the current fee rate for that specific account, not the network-wide fee rate. A transaction touching only uncontested accounts needs a much lower priority fee to achieve the same certainty of inclusion.

The implication for DApp developers is that blanket priority fee strategies no longer work well. An application that sets a fixed priority fee for all transactions — a common pattern before SIMD-0096 — will either overpay when accessing uncontested accounts or underpay when accessing contested accounts. The correct approach is to query fee estimates at the account level before each transaction, which requires infrastructure investment that many smaller DApps have not yet made.

Priority fee estimation services are updating their APIs to expose account-level fee data, but the ecosystem-wide tooling upgrade has a long tail. DApps that have not updated their fee estimation logic are, in the months following SIMD-0096 activation, either systematically overpaying or experiencing higher transaction failure rates on contested accounts than their users experienced before the fee market change.

MEV Under Local Fee Markets: What Changes

Maximal extractable value — the profit available to validators and searchers from reordering, inserting, or censoring transactions — is affected by local fee markets in ways that are not uniformly positive for the broader ecosystem.

Under the global fee market, MEV extraction was relatively blunt: searchers who wanted to front-run or sandwich a large DEX trade competed by paying high global priority fees, which raised the cost of all transactions during high-MEV events. Local fee markets change the structure: searchers now need to pay fees calibrated to the specific accounts involved in the target transaction, which in principle should reduce the collateral damage of MEV events on unrelated transactions.

In practice, the relationship between local fee markets and MEV is more complex. The introduction of account-level fee data creates new information that sophisticated searchers can use to identify high-activity accounts before submitting their own transactions — the congestion signal itself becomes an alpha signal for MEV extraction. Accounts showing elevated local fee rates are, by definition, accounts with high transaction demand, which is a proxy for accounts with high value flows worth capturing.

Jito, the Solana MEV infrastructure provider, operates a system of validator tip markets alongside the protocol fee market. Under SIMD-0096, the interaction between Jito tips and local fees creates a two-dimensional fee optimisation problem for searchers: they need to calibrate both the protocol-level priority fee for the specific accounts they’re touching and the Jito tip for block leader prioritisation. The equilibrium pricing of this two-dimensional market is still being discovered; strategies that worked well in the pre-SIMD-0096 environment are being recalibrated.

Validator Economics: Winners and Adjustments Required

Validators benefit from local fee markets in a specific way: fee revenue becomes more accurately correlated with the value of the block space being allocated. Under the global fee market, validators received elevated fees during any network congestion event; under local fee markets, they receive elevated fees specifically when the accounts that are generating the congestion are included in blocks. This more accurate matching of fee revenue to resource consumption should improve the long-run economics of block space allocation.

The near-term adjustment for validators is in scheduling. Solana’s banking stage — the component that accepts and sequences transactions from the incoming transaction pool — needs to correctly implement account-level fee prioritisation rather than global fee prioritisation. Validator client software has been updated to implement SIMD-0096, but the quality of that implementation varies across the validator set. Validators running older software or software that implements the local fee market logic imperfectly may be leaving fee revenue on the table or accepting transactions in a suboptimal order.

The competitive dynamic in the validator set means that validators running optimal SIMD-0096 implementations will, at the margin, capture more of the available fee revenue than validators running suboptimal implementations — creating selection pressure toward correct implementation over time. But in the months following activation, fee capture efficiency is unevenly distributed, which is a source of revenue variance for validators with the same hardware and stake weight.

How This Compares to Ethereum’s EIP-1559

The natural comparison point for Solana’s local fee markets is Ethereum’s EIP-1559, which replaced Ethereum’s first-price auction fee mechanism with a base fee plus tip structure in August 2021. The comparison is instructive but imprecise.

EIP-1559 addressed a different problem: Ethereum’s first-price auction created fee estimation uncertainty (users couldn’t predict what fee was required for inclusion) and volatile fee spikes during congestion. EIP-1559 introduced a protocol-level base fee that adjusts predictably based on block utilisation, with a tip added for prioritisation within the block. The base fee is burned, removing it from miner revenue and adding a deflationary mechanism to ETH supply.

Solana’s local fee markets address congestion localisation rather than fee predictability per se. Solana’s fee structure has always had a base fee component; the change is in how priority fees are scoped to resource contention. There is no equivalent to EIP-1559’s base fee burn mechanism in SIMD-0096 — the fee revenue, including priority fees, goes to validators and to the network’s burn mechanism at existing proportions. The comparison to Ethereum is therefore partial: both systems moved toward more market-efficient fee allocation, but the specific problems they solved and the mechanisms they used differ substantially.

What Solana and Ethereum share in the post-reform environment is a fee structure that requires application developers to do more work — querying dynamic fee data rather than using static fee parameters — in exchange for better user experience during congestion. The developer overhead is a real cost that smaller protocols may be slow to absorb.

What DApp Developers Should Do Now

For protocol and application developers on Solana, the practical response to SIMD-0096 has several components that are not all being implemented at the same rate across the ecosystem.

The immediate priority is updating priority fee estimation to use account-level data rather than network-wide data. The Helius, Triton, and QuickNode RPC providers have all published updated APIs that expose account-level fee estimates; the Solana SDK has been updated to support account-level priority fee queries directly. Applications that haven’t yet integrated these APIs are operating with fee estimation that is systematically miscalibrated under the new fee market architecture.

The second priority is transaction retry logic. Under the global fee market, a transaction that failed due to low fees could be resubmitted with a higher global priority fee. Under local fee markets, the retry strategy needs to account for whether the fee failure was due to account-level congestion or a different issue. Naive retry loops that simply increase the global priority fee on resubmission may not resolve local account congestion failures effectively.

The longer-term priority — relevant for protocols with complex cross-program invocations — is understanding the fee profile of each program the protocol touches and designing transaction flow to minimise exposure to contested accounts where possible. Some DeFi protocol architectures can be refactored to reduce the number of contested accounts touched per transaction; others have inherent account contention that cannot be architectured away and need to be managed through fee strategy instead.

What This Means for Solana’s Competitive Position

The local fee market implementation is a genuine network improvement that addresses a complaint that has followed Solana since its high-growth phase. The elimination of global fee spikes caused by unrelated activity removes one of the most common user friction points that drove transaction failures during peak demand. Silent churn from transaction friction is a measurable problem in crypto protocols, and reducing the frequency of inexplicable fee spikes during otherwise normal activity is a real product improvement.

The competitive significance depends on how well the ecosystem executes the transition. If the majority of Solana DApps update their fee estimation and retry logic within the next two to three months, the user experience improvement will be visible and attributable to the fee market change. If the ecosystem upgrade has a long tail — with many smaller apps running miscalibrated fee estimation for a year — the improvement will be uneven and the network’s reputation for reliability will remain mixed.

The comparison to Ethereum’s EIP-1559 rollout is instructive here too: the Ethereum ecosystem took six to twelve months after EIP-1559 activation for fee estimation across the application layer to be reliably calibrated. Solana’s faster-moving developer community may compress that timeline, but the challenge of coordinating a fee estimation upgrade across hundreds of independent protocols is real regardless of how capable the underlying developer community is.

FAQ

What is SIMD-0096?
SIMD-0096 is the Solana Improvement Document that implemented local fee markets on the Solana network. It replaced the single global fee rate with per-resource pricing, meaning transactions pay priority fees based on the congestion of the specific accounts and programs they access rather than network-wide congestion.

Why did Solana need local fee markets?
Solana’s original global fee market meant that congestion in any application — an NFT mint, a token launch — raised fees for every transaction on the network regardless of whether it touched the congested accounts. This created fee spikes and transaction failures for users whose activity was unrelated to the source of congestion.

What do DApp developers need to change?
Priority fee estimation needs to use account-level fee data rather than network-wide fee rates. Transaction retry logic needs to handle account-level fee failures correctly. Developers should update to current Solana SDK versions and integrate account-level fee APIs from RPC providers like Helius, Triton, or QuickNode.

How does SIMD-0096 affect MEV?
MEV extraction becomes more account-specific: fee spikes from high-MEV events are localised to the contested accounts rather than network-wide. However, account-level congestion data also provides new alpha signals for searchers. The two-dimensional fee optimisation (protocol priority fee + Jito tip) creates a more complex MEV extraction environment that is still being calibrated.

How does this compare to Ethereum’s EIP-1559?
Both reforms moved toward more market-efficient fee allocation, but they solved different problems. EIP-1559 introduced predictable base fees that adjust to block utilisation and burn the base fee. SIMD-0096 localises priority fees to resource contention without a base fee burn mechanism. The developer overhead — adapting applications to dynamic fee data — is similar in both cases.

Sources

• Solana Foundation — SIMD-0096: Local Fee Markets specification
• Helius — Priority Fees: Understanding Solana’s Updated Transaction Fee Mechanics
• Jito — Local Fee Markets and MEV: Implications for Searchers and Validators
• Solana Developer Docs — How to use priority fees under local fee markets
• Blockworks — Solana’s local fee markets go live: how the ecosystem is responding

What Local Fee Markets Mean for Platform Economics

The platform economics of local fee markets require thinking about two user constituencies whose interests are structurally opposed. High-frequency traders and searchers benefit from low baseline costs across high transaction volumes — they are the users whose extraction the MEV literature focuses on. Application developers and their end users need execution reliability: the guarantee that a transaction submitted with an appropriate fee will actually settle in the current slot rather than queue behind a wave of arbitrage. SIMD-0096 addresses the second constituency’s problem more directly than the first: by localising congestion to individual hot accounts rather than allowing it to propagate across the validator set, it restores execution reliability for the applications that drive the network’s addressable market beyond financial speculation. This is the design trade-off that differentiates platforms that sustain diverse use cases from chains that optimise for a single cohort of power users and then discover the revenue ceiling that monoculture produces.