AI Is Taking the Capital. Blockchain Has to Prove What It Is For
More than half of the venture capital invested in Europe during the first half of 2026 went into artificial intelligence companies. Around $23 billion flowed into European AI start-ups, up from $10 billion a year earlier. Worldwide investment reached $388 billion over the same six months, nearly three times the amount recorded in the first half of 2025.
Fundraising has kept pace with the deployment. Accel has raised an $800 million fund for Europe and Israel, where AI already accounts for a large share of new companies entering its pipeline. Highland Europe has raised €1.1 billion for later-stage technology companies. Lovable added another $400 million in August at a valuation of $13.3 billion.
Web3 companies now compete for money inside a venture market that has found another infrastructure story. A few years ago, blockchain founders raised capital around new protocols, token economies, faster execution and the prospect of replacing parts of the financial system. Investors now have a much simpler test for infrastructure businesses: who uses it and what do they pay for?
AI companies have answers. Developers pay for models and APIs. Companies buy software that automates work. Data centres need chips and electricity. Defence, robotics and biotechnology companies are incorporating AI into products that venture investors once regarded as too capital-intensive or too slow to finance.
Blockchain infrastructure has reached the point where technical improvement alone carries less weight.
AI has taken over the infrastructure pitch
Accel says AI is entering defence, biotechnology, materials science and robotics, sectors that historically sat outside the normal software venture model. Software now plays a larger part in research, simulation, design and prototyping, giving venture investors another way into industries that once depended mostly on industrial capital.
The spending extends far beyond applications. Technology companies are putting billions into chips, data centres, power and networking. AI funding therefore reaches both software and the physical systems required to run it.
Blockchain made a similar argument during the previous technology cycle. New Layer 1 networks promised more capacity. Layer 2 systems reduced transaction costs. Developers built decentralised storage, interoperability protocols and alternative execution environments. Investors financed much of the work before demand had caught up with the infrastructure.
Many of those systems now function well. The commercial test has become harder.
Another blockchain offering faster transactions needs enough applications to use them. A storage protocol needs customers prepared to move workloads away from established cloud providers. A new execution layer needs developers who have a reason to leave existing networks. Lower fees or higher throughput do not answer those questions by themselves.
AI companies are competing for capital with businesses that already show where demand comes from.
AI agents give blockchain a more concrete job
The overlap between AI and crypto becomes more interesting when software begins acting on its own.
An AI agent can search for a supplier, compare prices, use an API, rent computing capacity or arrange a purchase. Payment networks, banks and technology companies are already building systems around agentic commerce. Visa, for example, is developing payment infrastructure designed to let software agents transact under rules set by the account owner.
Conventional financial infrastructure handles much of that activity perfectly well. A card network processes a payment. A bank account holds money. A technology platform stores permissions and identity data. Adding a blockchain to an ordinary purchase rarely improves the experience enough to justify the extra infrastructure.
Transactions between independent software systems create a harder problem.
An agent buying services from a provider it has never dealt with needs clear spending authority. The provider needs to know whether payment will settle. Both parties may need a record of what the agent was authorised to buy, which software initiated the transaction and what conditions applied.
Programmable wallets already handle some of those controls. Smart contracts enforce spending rules or release funds after agreed conditions have been met. Stablecoins move across borders without requiring each software agent to maintain a banking relationship with every company it pays.
The argument for blockchain therefore becomes narrower and more practical. Software does not need decentralisation as an ideology. It needs a payment and settlement system that works across companies, countries and platforms without a person approving every transaction.
Machine payments change the economics
Human consumers already have cards, instant payments and mobile wallets. Most do not care what infrastructure sits behind a €20 purchase as long as the payment works.
Software generates a different pattern of transactions. One agent might purchase an API request worth a fraction of a cent, pay for data from another system, rent computing capacity for several seconds and settle with multiple providers during a single task. Traditional payment systems were not designed around millions of tiny machine-initiated purchases.
Stablecoins give software a digital cash instrument. Wallets provide programmable control over funds. Smart contracts automate conditions without requiring a person to return to a checkout page.
Crypto developers have spent years building those components. Agentic commerce now provides a reason to use them that does not depend on persuading consumers to replace cards or bank transfers.
Competition remains strong. Banks and payment companies already control customer relationships, compliance systems and access to fiat money. Visa does not need merchants to adopt a new financial network before adding credentials for AI agents. Banks developing programmable deposits start with regulated balance sheets and existing corporate customers.
Blockchain has an advantage where machines transact across organisations that do not share the same platform or banking infrastructure. The size of that market remains uncertain, but the commercial question is finally more specific than whether blockchain will disrupt finance.
More blockchain usage does not guarantee another blockchain boom
AI adoption may increase blockchain transactions without producing a broad new funding cycle for crypto start-ups.
An autonomous agent does not care which venture firm financed the underlying network. It needs predictable fees, reliable settlement and interfaces that developers can integrate easily. Existing blockchains and stablecoins already provide much of that.
Machine payments could therefore reinforce a small number of networks rather than create demand for dozens of new ones. Developers would have little reason to move to another blockchain offering slightly faster execution unless the new network solved a specific problem.
AI infrastructure already shows how quickly activity can concentrate. Thousands of companies build applications on top of computing infrastructure controlled by a limited number of cloud and chip providers. Rapid growth at the application layer has not produced an equally fragmented infrastructure market.
Crypto may follow a similar pattern. More agent transactions could mean heavier use of a few settlement networks, a small number of liquid stablecoins and wallet infrastructure that developers barely notice once integrated.
For Web3 investors, higher blockchain usage and higher returns from new blockchain infrastructure are therefore two different propositions.
Venture capital is applying a stricter test
Europe’s 2026 funding numbers show where investors currently see demand. AI companies received roughly $23 billion in the first half of the year while companies across every other sector shared around $19 billion.
Public markets have already questioned the pace of spending. AI chip and cloud companies lost more than $1 trillion in market value during one week in July before recovering. Investors have become more sensitive to the gap between money spent on AI infrastructure and profits earned from deploying it.
Venture funds continue putting money into the sector because early-stage investments run over longer periods and AI companies keep opening new markets. Accel is raising $3.5 billion across funds covering Europe, the US, India and follow-on investments.
A correction in AI valuations would not automatically send those billions back into crypto. Blockchain companies still have to compete for the same engineers, founders and investor attention against businesses selling products into markets where customers already spend money.
Agentic commerce gives parts of Web3 a credible place in that competition. Stablecoin settlement, programmable wallets, machine permissions and smart contracts address practical problems once software begins spending money independently across platforms.
The opportunity does not rescue every protocol. It puts pressure on founders to identify the exact transaction their infrastructure handles better than a bank, card network, database or existing blockchain.
Investors have spent years funding crypto infrastructure in anticipation of demand. AI is now forcing the sector to show where that demand actually exists.
