Paradigm: When AI Fund and Cryptocurrencies Meet in the Age of Artificial Intelligence

At the beginning of 2026, Paradigm announced a new fund of up to $1.5 billion. At first glance, it seems just another investment firm opening resources to explore opportunities in advanced technologies. But this move reveals something much bigger: a strategic repositioning of how the leading investors in the cryptocurrency sector view the future of tech venture capital.

The story behind this figure didn’t start in February 2026. It began with an accounting situation that increasingly highlighted a dilemma Paradigm itself needed to solve: owning $12.7 billion in assets under management but raising ever smaller funds. In 2021, Paradigm raised $2.5 billion. Three years later, in 2024, its fund had shrunk to $850 million. The issue wasn’t reputation or fundraising capacity. It was pure mathematics: there weren’t enough valuable crypto projects to absorb that volume.

The cryptocurrency market reached a saturation point

In 2025, while the total invested in crypto startups reached $49.8 billion globally, the number of venture capital deals in the sector fell by approximately 60%. It went from about 2,900 deals annually to just 1,200. The interpretation is paradoxical: more money, but for fewer projects. Capital is increasingly concentrated in a few large-scale operations, leaving little room for the traditional diversified venture capital model.

For a small fund, this may not be threatening. But for Paradigm, managing a portfolio of tens of billions, it’s a structural constraint. Maintaining the expected return quality while deploying billions in a contracting market has become progressively challenging. The very reduction in fund sizes was a strategic adaptation, albeit a necessary one.

From the chaos of FTX to the quiet restructuring

November 2022 marked a turning point. The collapse of FTX, Sam Bankman-Fried’s empire, shattered the trust that had been built. Paradigm’s recorded investment in that exchange was $278 million. It all evaporated to zero. For an institution positioning itself as a thought leader in the sector, based on rigorous technological research, it was a blow that went beyond numbers: a public failure of judgment.

What followed was significant precisely because of its subtlety. In 2023, observers noticed that Paradigm’s corporate website had undergone changes. Mentions of “crypto” and “Web3” were quietly removed, replaced by broader language about “technology investment.” It wasn’t an official announcement. The community discovered it, and the reaction sparked intense debate. The emerging question was inevitable: was Paradigm abandoning the sector?

Matt Huang, the firm’s co-founder, had to intervene publicly. He stated they were never “more excited about cryptocurrencies,” but added something essential: “AI development is so remarkable that it cannot be ignored. Positioning AI and crypto as zero-sum competition is a popular but inaccurate narrative. We disagree. Both are deeply interesting, and there will be significant overlap.” This statement was both a defense and a hint. It revealed that internally, Paradigm had already been seriously contemplating artificial intelligence.

The response Matt Huang has been building for two years

If someone only looks at Paradigm’s recent announcement, they might think this transformation is recent. But analyzing Huang’s concrete actions over the past 24 months makes it clear he was operating on a different level. He was no longer solely an investor in crypto.

In 2024, Paradigm invested $50 million in Nous Research, a company specializing in AI infrastructure and large open-source language model development. It wasn’t exploratory or marginal; fifty million represents a genuine institutional commitment. Simultaneously, in February of this year, Paradigm collaborated with OpenAI to launch EVMbench, a benchmarking tool designed to evaluate how different AI models detect and fix vulnerabilities in smart contracts. Here, the fundamental infrastructure of cryptocurrencies intersected with AI evaluation capabilities, bringing the two worlds to the same table.

Meanwhile, Huang co-founded Tempo, a stablecoin payment infrastructure. This participation isn’t casual: it aligns perfectly with his role on Stripe’s advisory board, which formed a strategic partnership with Paradigm in 2025 and launched its own stablecoin payment products that same year.

Putting these actions together, it’s clear: Huang wasn’t “about to invest in AI.” He had already been inhabiting the intersection of AI and cryptocurrencies for at least two years. His bet wasn’t solely on AI or on cryptocurrencies, but on the precise moment when these two forces would converge.

Why bet on the AI×Crypto fusion instead of simply shifting to AI

Paradigm’s expansion into AI doesn’t mean direct competition with giants like a16z or Sequoia for the same projects. A common misinterpretation would be to see this move as “another traditional VC discovering AI.” If that were the case, it would offer no advantage, since the broader AI space is already filled with larger funds with deeper resources.

The real logic is different: Paradigm isn’t competing for the entire AI pie. Its bet is on the intersection that most haven’t fully perceived yet. Autonomous AI agents are among the hottest concepts right now. These intelligent systems are beginning to replace human intervention in searches, coding, data analysis, and workflow management. But there’s an unresolved gap: money.

When an AI agent needs to make a payment, receive funds, or transfer resources between different services, what system does it use? PayPal? Traditional bank accounts? These systems were designed for humans, requiring manual authentication and authorization, incompatible with autonomous machine execution.

Stablecoins, on the other hand, can. Smart contracts can. Programmable money can. That’s precisely why Huang is developing Tempo (stablecoin payments) and investing in Nous Research (AI infrastructure). He bets that these two lines will converge, and Paradigm is positioned to capture enormous returns at that moment. It’s not a transformation; it’s a smart expansion into a territory others haven’t fully mapped yet.

The fund as a response to institutional investor expectations

There’s also a fundamental practical dimension. Paradigm’s LPs—institutions and individuals trusting their capital—saw their fundraising shrink from $2.5 billion in 2021 to $850 million in 2024. This significant contraction required a justification. A new $1.5 billion fund demands an even more compelling narrative.

“Continuing to invest in early-stage crypto projects”—that story became insufficient to sustain such a large raise in 2024 and 2025. But “leveraging established technological advantages in the crypto ecosystem to enter cutting-edge AI and robotics technologies”—that’s a narrative that works.

In 2025, 61% of all global venture capital capital flowed into AI, totaling approximately $258.7 billion. It’s currently the largest reservoir in the venture capital universe. The $1.5 billion fund Paradigm raises aims to draw water from this seemingly inexhaustible reservoir, not to remain stuck in a shrinking lake. For LPs, it’s a broader narrative with a more solid growth logic.

Returning to 2023: when Huang was compelled to clarify the website’s restructuring, he said, “AI and crypto are not a zero-sum competition.” At the time, it seemed like tactical defense—to calm the community, prevent investor panic, and leave room for AI exploration. But rereading it today reveals a preview of what was to come.

Paradigm spent three years recovering from the FTX ruins. Instead of choosing the safer path—shrinking scale, focusing solely on crypto, waiting for the next bull cycle—it opted for a more complex route, but with much greater expansion potential: building a presence in both sectors simultaneously, betting on the convergence of AI and crypto, waiting for the moment these forces meet. The $1.5 billion fund is simply a marker of this stage of the journey.

Huang did not publicly respond to the Wall Street Journal report on February 28. But his Tempo continues to develop, his involvement with Nous Research remains active, and EVMbench is already operational. Additional explanations are unnecessary. Actions have already communicated much more than any statement ever could.