Billionaire Kevin O’Leary shouts "The next step in the AI wave is web3": LLMs cannot create Starbucks, but Blockchain can.

Kevin O’Leary pointed out that AI and blockchain will reshape retail, but on-chain scalability remains the biggest challenge. The market is seeking breakthroughs with DAG and Agentic protocols. (Background: Rich Dad claims “Ethereum and silver” are the best assets currently: the global economy will collapse this year, and paper assets will become worthless.) (Supplementary background: Warren Buffett, who has always been skeptical of gold, has quietly bought in? Rich Dad: This means a market collapse is not far off.) Venture capitalist and well-known billionaire Kevin O’Leary made a prediction at a public event on the 19th: in the future, consumers will only need to say one sentence, and AI agents will be able to complete store selection and ordering, with blockchain synchronously settling the payment, all in less than two minutes without needing to take out a phone. This statement immediately sparked discussions on the X platform, as it connects the two technological waves of AI and blockchain to the most sensitive pain points in retail—speed, cost, and experience. AI automatic ordering, blockchain frictionless payment. O’Leary painted a vivid picture: “I want a fat-free latte, and it will arrive in 90 seconds.” Through so-called “Agentic AI (,” the system first reads the user's location and preferences, then compares the inventory of nearby retailers, queue times, and delivery routes, ultimately sending the order to the blockchain payment layer for deduction. For retailers, the benefit is more precise demand forecasting and instant payment; for consumers, everything from price comparison, ordering to payment is automated. Huge traffic challenges blockchain scalability. Behind this ambitious vision, the biggest shadow comes from blockchain scalability. O’Leary likens current mainstream chains to “highway toll booths”: they get congested when there are too many vehicles. Taking Ethereum's fees and processing speed as an example, transaction costs during peak times may far exceed the price of a regular coffee, making it impossible to support the daily transaction volume of Walmart-level sales. Retail payments need to achieve millisecond-level confirmation, nearly zero fees, and stable operation around the clock; the current on-chain structure is still noticeably far from this threshold. DAG and ACP: Potential breakthrough paths. The market has not stopped at complaints. On the underlying level, developers are turning their attention to directed acyclic graphs )DAG(. DAG uses a “web-like” layout, allowing multiple transactions to validate each other simultaneously without waiting in line for packaging, theoretically achieving TPS in the tens of thousands. Some startups like BlockDAG, Hedera, and Nano have commercialized this structure, intending to enter the micro high-frequency market. On the application level, Agentic AI is waiting for standardization. Stripe and OpenAI have jointly proposed the Agentic Commerce Protocol )ACP(; Google’s AP2 is also making progress in the same direction. These protocols attempt to define how AI agents securely access wallets, verify identities, and process refunds, enabling large retailers and payment providers to connect. Risks, opportunities, and timelines. Research institutions predict that the market size of blockchain in retail could reach $6.01 billion by 2030, while more than half of generative AI companies will deploy autonomous agents by 2027. This means that if infrastructure can break through in the next two to three years, early participants can directly reap double dividends. Risks also exist: identity verification, user control over funds, and regulatory challenges are still being explored. Even if technology matures, large retailers will need time to transform their cash register systems and backend ERP. For investors, the threshold is both a resistance and a moat. DAG projects, wallet service providers, and payment API suppliers are all on the same starting line, competing for the key nodes of “on-chain millisecond settlement” and “AI agent dispatch.” Once the iteration speed is validated, consumers will only need a single command, and the retail process can hand over everything from decision-making to payment to algorithms, just as O’Leary envisioned. O’Leary speaks casually about the future, but to make “90-second ordering” a reality, engineers still face the cold reality of blockchain throughput and regulatory grey areas. Nevertheless, the parallel processing of DAG structures and the standardization of protocols like ACP have opened up feasible paths for this revolution. As the benefits of AI agents align with efficient on-chain settlement, a new tipping point in retail will emerge—at that moment, ordering a cup of coffee could be faster than opening a wallet. Related reports: Testing three major exchanges' AI assistants with “1011 big dump”: Hexagonal player, Riddler, and one with “Android thinking”. Suspected “1011 insider whale” turns from fall to rise: launching $7.6 million in BTC and $35 million in ETH long orders. After the “1011” epic liquidation in the crypto market, are DAT companies' stocks still doing well? <Billionaire Kevin O’Leary says “the next step under the AI wave is web3”: LLM cannot create Starbucks, but blockchain can> This article was first published in BlockTempo, the most influential blockchain news media.