When global capital markets are still debating whether an AI valuation bubble has formed, some of the most perceptive investors have already placed their bets on the upstream "pick-and-shovel" providers. On May 4, 2026, Haun Ventures—founded by former a16z partner and federal prosecutor Katie Haun—announced the completion of a new $1 billion fundraising round, officially expanding its investment scope from blockchain infrastructure into AI agents and the intelligent economy.
This move is not an isolated event. Just one day after Haun Ventures’ announcement, a16z crypto closed its fifth crypto fund, totaling $2.2 billion. Both top-tier VCs are now focusing on the intersection of AI and crypto. However, while a16z pursues a broad strategy of "turning infrastructure into everyday products," Haun Ventures takes a more targeted approach. It explicitly lists "agent economy," tokenized assets, and crypto financial infrastructure as its three core investment pillars, emphasizing that AI investments must remain "in their lane"—meaning it will only invest in AI projects that directly intersect with crypto infrastructure, not in generic AI models or application layers.
What underpins this investment logic? The answer points to a rapidly forming consensus: as competition at the AI model layer intensifies and training costs approach hundreds of millions of dollars, structural shortages in compute supply have become the industry’s biggest bottleneck. Decentralized GPU compute networks—led by the Render Network—are positioned at the forefront of this gap.
Paradigm Shift: From "Crypto Fund" to "AI+Crypto Infrastructure Fund"
Haun Ventures’ latest fundraising is not a sudden pivot, but the result of a multi-year strategic evolution.
The firm debuted in 2022 with a $1.5 billion inaugural fund, marking the tail end of the previous crypto bull cycle and setting a record for female-founded VC fundraising. However, just months after its launch, the collapse of FTX plunged the entire industry into a deep winter. Haun Ventures adopted an extremely cautious deployment pace—by mid-2023, about 60% of its first fund’s capital remained unused.
This "holding pattern" period actually laid the groundwork for its current strategic shift. During this time, three structural changes gradually emerged:
First, AI compute demand entered an exponential growth trajectory. NVIDIA CEO Jensen Huang stated at CES 2026 that AI compute needs are "growing by orders of magnitude every year." Gartner projects global AI spending will reach $2.52 trillion in 2026, up 44% year-over-year, with AI infrastructure contributing an additional $401 billion.
Second, the crypto industry’s infrastructure narrative evolved from "trading tools" to "economic rails." In 2025, annual stablecoin transaction volumes surpassed the multi-trillion mark, rivaling mainstream card networks. This provides a practical settlement layer for on-chain economic activity by AI agents.
Third, decentralized physical infrastructure networks, after an early phase driven by token incentives and "empty cycles," finally found genuine paying demand amid the AI compute boom. In 2025, the top three DePIN projects by revenue all focused on GPU compute sales, moving away from storage, bandwidth, or sensor data.
The convergence of these three shifts forms a logical loop for Haun Ventures’ $1 billion fund: crypto financial infrastructure, tokenization, and AI agents. It’s important to clarify that this fund isn’t solely for AI or crypto, but specifically targets the infrastructure layer at their intersection.
The True Scale of the Compute Supply-Demand Gap
To understand the value of Haun Ventures’ strategic pivot, start with a core figure: just how tight is global AI compute?
Bridgewater Associates forecasts that in 2026 alone, major US tech firms will invest about $650 billion in AI infrastructure. Meanwhile, global GPU infrastructure spending is expected to surge from $83 billion in 2025 to $353 billion in 2030, with AI compute demand growing at 37% annually.
Yet, supply expansion lags far behind demand. SK Hynix and Micron, two leading high-bandwidth memory (HBM) producers, have announced their entire 2026 capacity is sold out. Samsung faces similar constraints, and all three major HBM suppliers have fully booked their production. This supply bottleneck is creating a "two-tier market": top AI labs like OpenAI and Anthropic lock in GPU resources at near-cost prices via multi-billion-dollar "equity-for-compute" deals, while smaller firms without strategic partnerships are forced to pay retail prices several times higher.
This structural inequality in compute allocation is the fundamental demand driver for decentralized GPU networks. Two-thirds of global cloud compute is controlled by AWS, Azure, and Google Cloud, meaning most AI developers and startups face not just cost issues, but access barriers.
There is a significant global GPU compute gap, with top firms and smaller players facing cost disparities of several multiples. For 2026, HBM memory capacity is already locked by major suppliers, confirming a sold-out status. Decentralized compute networks, by aggregating idle GPU resources, can theoretically offer elastic supply at prices well below centralized cloud providers, but earning enterprise trust remains a challenge. If the current GPU supply bottleneck persists into 2027, decentralized compute networks may see a critical window for enterprise adoption.
Sector Breakdown: Who’s Capturing This "Certain Demand"?
Within the decentralized compute sector, Render Network offers the most complete narrative, but it’s not the only player. Understanding the competitive landscape helps evaluate Haun Ventures’ chosen industry direction.
Render Network originally focused on decentralized GPU rendering—connecting node operators with idle GPUs to 3D artists and VFX studios needing rendering power. Its core engine, OctaneRender, and partnerships with Apple, Microsoft, Google, and NVIDIA provide industry credibility unmatched by peers.
What truly brought Render into the AI spotlight were a series of strategic moves from late 2025 to early 2026:
First, the launch of the Dispersed.com platform in December 2025 marked Render’s formal expansion from 3D rendering into general-purpose AI compute. The platform aggregates decentralized GPUs for AI model training and inference, already integrating enterprise-grade NVIDIA H200 and AMD MI300X GPUs.
Second, in April 2026, community voting approved the RNP-023 proposal, bringing Salad’s decentralized subnet into the Render ecosystem exclusively. Salad previously operated the world’s largest consumer GPU network—about 60,000 daily active machines across 180+ countries. This integration fundamentally changes Render’s supply structure, expanding from professional nodes to consumer GPUs and greatly enhancing multi-scenario coverage.
Third, Render uses a burn-mint equilibrium model, where a portion of network usage fees is destroyed. According to RenderCon 2026, AI workloads now account for roughly 35%-40% of network usage.
Gate market data as of May 6, 2026 shows Render Network’s token RENDER trading at $1.90, with a 24-hour volume of $576,900, up 3.68% in the past 24 hours. Circulating market cap is $983.9 million, circulating supply is 518.74M tokens, max supply is 532.21M, and circulating market cap is 97.47% of fully diluted value. The token is up 7.79% over the past 7 days, but down 56.69% over the past year. Current market price remains well below its all-time high of $13.59.
In 2026, Render Network completed its strategic shift from 3D rendering to AI compute, significantly expanding supply through the Dispersed platform and RNP-023 proposal. Render’s AI compute pivot is compelling at the narrative level, but whether it can surpass traditional rendering revenue remains to be seen. If Salad’s 60,000 GPUs are successfully integrated and achieve high utilization, Render’s burn-mint equilibrium may show stronger deflationary effects in the second half of 2026.
In contrast, the ASI Alliance—formed by Fetch.ai, SingularityNET, and Ocean Protocol—does not provide compute itself, but aims to build decentralized general AI infrastructure: agent coordination, cross-chain operations, and data marketplaces. The key milestone for 2026 is the final 1:1 migration from FET to ASI tokens. If Render’s business is "renting out GPUs," the ASI Alliance is building "economic rails for on-chain agent trading and collaboration." Both occupy distinct niches within the same overarching trend.
Market Sentiment Breakdown: Optimists vs. Skeptics
Market opinion on decentralized compute and Haun Ventures’ bet is sharply divided.
Optimists focus on three points: first, the structural growth in AI compute demand is highly certain—NVIDIA’s data center business continues rapid growth, with annual revenue exceeding $130 billion, underscoring genuine demand; second, decentralized compute offers disruptive cost advantages—decentralized GPU networks can provide batch and inference compute at 60%-80% lower prices than traditional cloud providers, with even greater advantages in some scenarios; third, Haun Ventures’ previous exits add credibility—Stripe acquired its stablecoin platform Bridge for $1.1 billion, Mastercard bought BVNK for up to $1.8 billion, and Katie Haun’s deep expertise in crypto compliance and policy is seen as a strategic asset.
Skeptics raise several concerns: first, crypto VC is in a contraction phase. In April 2026, total crypto fundraising plunged 74% to $662 million, a 12-month low, with large rounds "disappearing entirely." Whether Haun Ventures’ countercyclical raise has sufficient market foundation remains to be seen; second, according to CoinDesk, Digital Asset CEO Yuval Rooz notes a "huge gap between expectations and actual business volume"—flagship decentralized compute projects still lag traditional cloud providers by orders of magnitude in revenue and paying users; third, compliance, security, and service quality certifications of commercial cloud platforms are not easily replaced by decentralized alternatives in the short term.
This debate boils down to normal market dynamics between "long-term logic" and "short-term validation." Haun Ventures’ $1 billion bet is on the former’s certainty; skeptics are waiting for the latter’s proof.
Industry Impact: Structural Establishment of a New Sector
Haun Ventures’ fundraising and a16z’s follow-up together send a clear signal: the infrastructure intersection of AI and crypto has moved from "edge narrative" to "core allocation" for top VCs.
Notably, Haun Ventures achieved countercyclical expansion amid overall crypto VC contraction. According to Fortune citing SEC filings, leading firms like Paradigm, Pantera, and a16z crypto saw AUM declines in 2025, while Haun Ventures’ AUM rose from $1 billion to $2.5 billion. This contrast indicates a growing institutional consensus to treat "AI+crypto infrastructure" as a distinct sector, not just a conceptual discussion.
Industry participants can draw three main takeaways:
First, VC allocation logic is changing. When VCs view AI and crypto as "converging technologies" rather than "two parallel tracks," teams capable of mastering both fields’ complexities gain asymmetric competitive advantages.
Second, structural opportunities for startups. Haun Ventures’ targeted investments in "picks and shovels"—payment rails, custody and identity systems, tokenization platforms—are all enablement layers. This suggests platform-level opportunities may be more valuable than application-layer plays in the AI-crypto intersection.
Third, implications for ordinary participants. The DePIN sector is no longer just a "narrative game" driven by token incentives; AI compute demand provides real paying scenarios and revenue models. However, moving from "paying scenarios" to "sustainable business models" still requires overcoming hurdles in technical stability, enterprise trust, and compliance frameworks.
Scenario Analysis: Three Possible Paths
Based on current public information and industry trends, the following scenario projections outline potential futures for decentralized compute infrastructure. Note: all scenarios are hypothetical and logical extrapolations, not predictions of market performance or asset prices.
Path One | Persistent Compute Shortage, Decentralized Networks Accelerate Adoption
Assumptions: NVIDIA supply chain bottlenecks persist, HBM memory capacity expansion fails to meet explosive AI inference demand. HBM capacity lock-in continues through 2027, and global GPU compute gaps remain significant.
In this scenario, small and mid-sized AI firms and independent developers face continued "compute famine," forcing them to seek alternatives outside centralized clouds. Projects like Render and Akash may achieve substantial breakthroughs in enterprise GPU onboarding and hybrid compute architectures, with network revenues potentially reaching hundreds of millions. Haun Ventures’ investments in AI agent financial infrastructure will directly benefit from increased on-chain agent transaction volumes.
Path Two | GPU Supply Recovers, Decentralized Network Cost Advantage Narrows
Assumptions: NVIDIA and AMD expand capacity successfully, HBM bottlenecks ease. GPU instance prices on AWS and Azure fall significantly.
Here, the core value proposition of decentralized compute—low cost—faces compression. However, irreplaceable advantages remain: elastic supply, no long-term lock-in contracts, and compliance flexibility from decentralized data. Competition shifts from "price wars" to "service quality" and "enterprise trust-building."
Path Three | AI Agent Economy Booms, Compute Demand Shifts to Inference, Decentralized Networks Gain Structural Opportunity
Assumptions: By late 2026, 40% of enterprise applications deploy task-specific AI agents (Gartner’s current forecast), and agent-to-agent transactions surge.
In this scenario, compute demand shifts from "training" to "inference," driving needs for low-latency, geographically distributed, on-demand GPU supply. Decentralized networks’ natural advantages—global node distribution and elastic scheduling—may unlock market space far beyond current expectations. Render and the ASI Alliance’s dual focus on hardware compute and agent economic rails could create synergistic effects between 2027 and 2028. Achieving this path depends on the pace of AI agent commercialization, which faces uncertainty in technology, regulation, and market demand.
Conclusion
Haun Ventures’ $1 billion fundraising is, at its core, a clear answer to the question: "Who benefits from the AI boom?" In Katie Haun’s framework, the answer isn’t the application layer or the model layer—it’s the infrastructure layer: networks that provide payment rails for AI agents, decentralized compute for training and inference, and tokenized channels for asset flow.
The brilliance of this positioning is that it doesn’t bet on any single AI industry winner, but on the irreversible infrastructure demand driven by AI’s continued expansion. No matter which AI model prevails, it will need compute for training and inference, on-chain economic rails for agent operations, and programmable asset layers for transaction settlement.
From a broader perspective, decentralized compute infrastructure is undergoing a critical shift from "crypto-native narrative" to "real AI industry demand." AI compute needs are giving the DePIN sector not just a new growth story, but a genuine paying customer base and real product-market fit.
Of course, this sector is far from being in a position to rest easy. Building enterprise trust takes time, validating revenue models requires data, and perfecting compliance frameworks demands policy negotiation. But the direction is clear—when the AI industry’s train accelerates by orders of magnitude, those laying the tracks will ultimately reap their rewards.
