The above button links to Coinbase. Yahoo Finance is not a broker-dealer or investment adviser and does not offer securities or cryptocurrencies for sale or facilitate trading. Coinbase pays us for certain activity generated through this link. Prices displayed are informational.

Investors have spent the last few years chasing AI growth by buying shares of companies like Nvidia and Microsoft. But a senior Wall Street executive says that strategy may not work for what comes next.

Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton (NYSE: BEN), an asset manager overseeing approximately $1.78 trillion as of May 31, argues in a LinkedIn post that agentic AI, not the generative AI most investors are familiar with, will be the technology that finally drives mainstream blockchain adoption.

Related: Solana's tokenized asset volume hits $5.8 billion in a record quarter

Generative AI, the kind behind tools like ChatGPT, mainly helps people create content and answer questions. Agentic AI is different. It refers to systems that can independently plan and carry out multi-step tasks, from initiating a transaction to tracking it through to completion, without a person guiding every step along the way.

Kaul cites estimates suggesting agentic commerce, meaning AI agents making purchases and payments on their own, could reach $3 trillion to $5 trillion by 2030.

That shift matters more than it sounds. As AI reshapes how companies spend money, the stakes are already visible. Kaul points to IBM's stock, which plunged 25.2% on July 14 after the company warned that customers were redirecting spending toward AI infrastructure and away from traditional software.

Analyst who called Bitcoin's drop from $100,000 to $75,000 now sees $50,000 as the next stop

Study finds most AI crypto trading agents aren't really trading

New Senate bill could ban Trump from launching his own tokens

For AI agents to transact with each other, autonomously and constantly, they need a payment system built for tiny, frequent transactions. A standard credit card charges roughly 2%-3% plus a flat fee near $0.30 per transaction.

Kaul notes that AI agent payments can cost as little as $0.001, for something like a single second of computing power. Traditional payment rails simply weren't built to handle transactions that small or that frequent.

Blockchains, Kaul argues, solve this naturally. They can verify a machine's identity cryptographically, record every transaction on a public ledger, and settle payments almost instantly.

She points to newer, faster blockchains like Aptos and Solana, which she says can process thousands of transactions per second, faster than networks like Bitcoin or Ethereum, and comparable to how quickly Visa's network operates day to day.

Kaul's broader argument is that capturing this shift will require more than owning shares of AI companies.

As AI agents increasingly transact directly on specific blockchains, she expects demand for those blockchains' native cryptocurrencies, and the smaller tokens built on top of them, to grow alongside that activity.

She compares the coming shift to the earlier jump from static websites to interactive, cloud-based apps, arguing that new leaders will emerge just as they did during that transition.

Related: Aptos now lets its biggest stakers run transactions at 100x the normal limit

This story was originally published by TheStreet on Jul 23, 2026, where it first appeared in the MARKETS section. Add TheStreet as a Preferred Source by clicking here.