Investors are directing more capital toward artificial intelligence, prompting fresh questions about the outlook for crypto. Industry veterans argue the two fields are converging rather than competing.
Payments evolution
When Greg Genega joined KPMG’s digital asset practice, blockchain was the rising star of technology that investors watched closely. Now, generative AI has absorbed much of that attention. Observers have begun asking whether crypto professionals are pivoting away from the field.
Genega, who has tracked digital assets through multiple cycles, pushes back on that framing. "It’s less of a pivot, more like a convergence," he told Sandmark.
Builders and investors are moving into AI at a rapid pace. Statistics from GitHub show that developers’ publishing of crypto-related code has declined by 75% even while the overall base of developers has grown. Much of that growth stems from increases in publishing of AI-related code.
Investment is following a similar pattern, with 80% of global venture funding flowing into AI in Q1 2026 according to Crunchbase. For those that are investing in crypto, 40 cents of every dollar is now being spent on AI-related projects. This backs up Genega’s observation on the convergence of the two technologies.
As token prices decline, the areas where the two technologies align may inform the evolution of the crypto industry. One immediate illustration of the touchpoints between AI and crypto is in payments.
AI agents – autonomous programmes capable of executing tasks and transactions independently – have posed new frameworks for businesses that have received extensive investment. By Q1 2026, 66% of technology functions and 55% of operations' functions were already deploying agentic AI, according to KPMG's first Global AI Pulse survey. Organizations are planning to invest an average of $186mn in AI over the next 12 months.
Agentic development has created crystallized ongoing challenges in traditional payments. Card networks carry fixed per-transaction fees that make sub-dollar payments impractical. Settlement also observes banking hours and is subject to the costs of crossing borders, none of which serves machine-to-machine commerce effectively.
Crossmint, a company building payment tools for AI agents, argues that stablecoins address each of these problems in ways existing rails fundamentally cannot. Stablecoins are cryptocurrencies designed to maintain stable value, typically pegged to the US dollar. "Paying an inference provider three cents or settling per-API-call needs a different cost structure," Alfonso Gómez-Jordana, co-founder of Crossmint, said. Stablecoins can move money at fractions of a cent, allowing for the micropayments necessary for agentic commerce.
He explained that the programmability of stablecoins is also important, allowing for specific rules to be set to limit and direct agents’ spending. "AI agents are going to be a defining part of this next era. As more economic activity gets initiated by agents rather than humans, the infrastructure has to serve them directly," he continued.
Verification layers
A limitation of AI development, however, is the risk of hallucination – where large language models generate plausible but incorrect information. While Genega noted that many of his clients are investing in AI development, the technology still needs a "human in the loop."
"Onchain provenance is one substrate that an LLM genuinely cannot hallucinate," he continued. Onchain refers to data recorded directly and immutably on a blockchain, making it deterministic and verifiable.
The verifiable nature of blockchain data also helps in the verification of deepfakes. As AI-generated content becomes indistinguishable from human-produced content, proving that a real person is on the other end of an interaction is becoming an equally urgent problem. "Cryptographic identity might become the only durable way to prove a human is on the other end of an interaction," Genega says.
He explained that the verification process can work both ways. While AI can hallucinate when building from scratch, it can be very effective when checking things like the coding of smart contracts. The same tools can also be used to check for vulnerabilities and alert for hacks in real-time.
Two halves
"They answer different halves of the same question," said Genega. "Whereas AI scales more of the creation of economic activity and crypto could scale its verification."
The development of AI could bring crypto more into the fold of mainstream finance. Once AI agents become routine economic actors transacting on blockchain rails, digital asset rails become the programmable layer that naturally intersects with that AI agentic world. The more economic activity AI creates, the more it needs the kind of trust infrastructure that crypto was built to provide.
"I don't think either technology, GenAI and blockchain, can live up to their full potential without the other one in existence," Genega concluded.
