Coinbase CEO Says AI Agents Will Need Crypto More Than Humans Do — and He Is Building the Infrastructure to Prove It

Coinbase CEO Brian Armstrong has a message for anyone who thinks crypto should step aside for artificial intelligence: the two technologies do not compete — they need each other. In fact, Armstrong argues, AI agents will eventually need crypto far more than humans ever did, because software programs cannot open bank accounts, wait three days for a wire transfer, or comply with Know Your Customer rules designed for people who carry passports.

By Amir Hassan | July 28, 2026

The Hook: Why the “Pivot to AI” Advice Misses the Point

In a series of posts on July 26, Armstrong pushed back against a narrative that has gained traction in tech circles over the past year: the idea that crypto companies should abandon digital assets and redirect their resources toward artificial intelligence. He called that framing “zero-sum, scarcity thinking” and laid out a case for why blockchain infrastructure is the missing layer that AI needs to function autonomously.

“Crypto is a general purpose technology,” Armstrong wrote. “It’s infrastructure, the same way electricity or the internet is infrastructure. It doesn’t compete with the next big thing, because it underpins it. It’s an and, not an or.”

The core of his argument is straightforward: AI agents — software programs that can make decisions and take actions on their own — need money to function in the real world. They need to pay for data, compute power, and services. They need to receive payments for tasks they complete. And they need to do all of this without human intervention. Traditional banking, with its branch-hours limitations, multi-day settlement times, and identity requirements, simply was not built for software that operates at machine speed across borders.

On-Chain Evidence: What Coinbase Is Actually Building

Armstrong is not just theorizing. He pointed to three specific Coinbase products that already power what he calls “agentic payments” — financial transactions carried out by AI programs rather than humans:

• Base — Coinbase’s layer-2 blockchain network built on Ethereum, which provides a fast and low-cost platform for smart contracts and decentralized applications. Think of it as an express lane on a highway: it runs alongside the main Ethereum network but processes transactions more quickly and cheaply.
• USDC — a stablecoin (a digital dollar) that maintains a 1-to-1 peg with the U.S. dollar, giving AI agents a stable unit of account that does not swing wildly in value like Bitcoin or Ethereum might.
• The x402 protocol — a payment standard developed by Coinbase that enables automated, real-time transactions between AI agents. Armstrong claims it already powers “the vast majority of all agentic payments” happening today.

Together, these three pieces form what amounts to a financial operating system for AI. An AI agent running on a server in one country can hold USDC in a wallet on Base, pay for cloud computing services through the x402 protocol, and receive payment for completed work — all without a human ever touching a bank account.

The Core Conflict: Can Traditional Finance Serve AI at All?

The tension Armstrong is highlighting goes beyond Coinbase’s product roadmap. It raises a genuine question about whether traditional financial infrastructure can adapt to a world where software programs are economic actors.

Consider the obstacles. A bank account requires a human identity — a Social Security number, a physical address, government-issued identification. AI agents have none of those things. Wire transfers take one to three business days to settle. AI agents operate in milliseconds. Banks operate within national borders and banking hours. AI agents run around the clock, globally, without regard for jurisdictions.

Armstrong’s prediction is that AI-driven transaction volume will eventually surpass what humans generate. “They will eventually transact far more per day than all humans combined,” he wrote. If that proves even partially correct, the demand for programmable, instant, borderless money — which is exactly what crypto provides — could scale dramatically.

But the skeptical counterpoint is worth considering. Crypto’s own infrastructure still faces real limitations: network congestion during high-usage periods, regulatory uncertainty in major markets, and user experience challenges that make it harder to use than a credit card. The vision of millions of AI agents seamlessly transacting on blockchain networks assumes these problems get solved — and that is not guaranteed.

Market Implications: What This Means for Your Portfolio

For regular investors, Armstrong’s thesis matters because it reframes the crypto investment case. If blockchain technology is not just about Bitcoin as digital gold or DeFi yield farming, but is instead the financial plumbing for the AI economy, then the total addressable market expands significantly.

With Bitcoin currently trading around $63,175 and Ethereum at approximately $1,875, the broader crypto market is pricing in a period of uncertainty — driven largely by macroeconomic concerns like the Federal Reserve’s rate decision due Wednesday and a semiconductor stock selloff that has dragged risk assets lower across the board. But the long-term investment case for blockchain infrastructure may be less about today’s price action and more about whether the technology becomes embedded in the next wave of digital commerce.

Armstrong envisions a future where AI agents do more than just make payments. They will trade assets, manage investment portfolios, raise capital for projects, and handle financial chores like tax planning and bill payments. He branded this emerging category “Agentic Finance,” or “AiFi” — a term that could become as common as “DeFi” if the vision materializes.

Investors watching this space should pay attention to which companies are actually building infrastructure that AI developers want to use. Coinbase’s Base network is one early leader, but the broader category includes any blockchain platform that offers fast, cheap, programmable transactions. The winners will be the networks that AI agents actually choose to transact on — not necessarily the ones with the best marketing.

The Verdict: Infrastructure Bets Are Long-Term Bets

Armstrong’s argument is compelling in its simplicity: AI needs money to function, and crypto is the only financial system built for software. Whether his timeline is right — and whether Coinbase captures the dominant share of this market — remains an open question. But the directionality is hard to argue against. As AI capabilities continue to advance, the demand for autonomous financial infrastructure will grow alongside them.

For investors, the takeaway is this: blockchain technology’s value proposition extends well beyond cryptocurrency prices. If the infrastructure being built today becomes the settlement layer for trillion-dollar AI-driven commerce, the networks and companies that establish early dominance could see sustained demand for years to come. That is a long-term thesis, not a short-term trade — and it requires patience through the kind of market volatility we are seeing today.

The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.