Buffett backs Google's $185B AI bet, calls race 'a game they don't want to play'

Warren Buffett has long avoided technology stocks, citing his inability to understand them. That changed with Berkshire Hathaway's investment in Alphabet, Google's parent company. In a CNBC interview, Buffett revealed that he personally initiated the $31 billion（約5兆円） stake—correcting the record after some had attributed it to incoming Berkshire CEO Greg Abel. 'I initiated it,' Buffett said. 'He's not doing anything I don't approve of. We talk all the time.'

Berkshire began building the position in the third quarter of 2025 and accelerated it this year, spending an additional $10 billion（約1.6兆円） in just the past month. The investment catapulted Alphabet shares nearly 4% on Wednesday, with the stock still climbing Thursday. Buffett said Alphabet now ranks 'number five or six' in Berkshire's holdings by value, and his endorsement added $8 billion（約1.3兆円） to Google co-founder Larry Page's net worth, pushing Page's personal wealth above $300 billion（約48兆円） for only the second time ever.

Buffett's rationale for the investment is that hyperscalers have fundamentally changed their business model. They are now spending hundreds of billions of dollars on capital-intensive infrastructure—data centers and chips—rather than relying on software licensing. This mirrors the business structure of railroads and utilities, sectors Buffett has long understood and invested in. 'The real question with Google and all of its competitors now, because they're all laying out hundreds of billions, and that's real money,' Buffett told CNBC. He added that the companies 'don't have any choice' in making this spending: they are forced to do so to remain competitive in the AI race.

Yet Buffett struck a cautionary note about the sustainability and rationality of this competition. He framed the AI capex race as 'a game they don't want to play,' comparing it to IBM's historic decline. 'IBM would have loved it if they just kept playing the game that IBM was playing in the 30s or the 40s or the 50s or the 60s,' he remarked, suggesting that while Google may be the strongest competitor in this new arena, the arena itself is a competitive burden. Despite this caveat, Buffett expressed rare confidence in Alphabet, saying Google is 'more likely to be a winner based on the record than probably 90% or 95% of what gets merchandised through Wall Street.' This assessment comes as Google doubles its AI spending to $185 billion（約30兆円）—a figure that even weighs on CEO Sundar Pichai. On Google's Q4 earnings call in February, Pichai acknowledged the pressure of converting those billions into tangible results and overcoming compute bottlenecks, though he expressed confidence in maintaining 'a very, very relentless innovation cadence' through 2026.

Buffett's investment marks a notable shift in his long-standing aversion to technology. He explicitly admitted, 'I made a mistake' in passing on Google for years, signaling that the company's business model has fundamentally changed in his eyes. The key insight from his CNBC remarks is that hyperscalers have moved from software licensing and digital services—domains Buffett found opaque—into massive physical capital deployment. This reframing allows him to apply the same disciplined logic he uses for railroads and utilities: understand the competitive moat, estimate the capex required, and assess whether the returns justify the spending.

Yet Buffett's characterization of the AI spending race as a game 'they don't want to play' is a crucial caveat. He invokes IBM's historical precedent—the company that dominated computing in earlier eras but was forced to compete in markets it did not choose—suggesting that while Google may be well-positioned to win this race, the race itself is a competitive necessity imposed by rivals, not an opportunity the company would freely elect. This nuance suggests Buffett sees Google as the strongest player in a structurally difficult situation, not as a company entering an attractive new market.