Berkshire to Use Narrow AI Only When It Adds Value

At Berkshire Hathaway’s annual meeting in Omaha on May 2, vice chairman Greg Abel outlined a policy to use narrow artificial intelligence only when it delivers clear operational benefits such as improved efficiency, safety or decision-making. Abel presented the approach in his first meeting as Warren Buffett’s designated successor.

Abel pointed to targeted applications already in use across the conglomerate. At railroad unit BNSF, he said specific analytics are helping to sharpen operations. Insurance businesses apply technology to screen claims for fraud and to detect synthetic media. The meeting opened with an AI-generated video of Buffett; Abel described that clip as a serious risk the company monitors daily and added, “It has to be additive to our businesses. We’re not going to do AI for the sake of AI.”

Buffett, who recently stepped down as chief executive, did not comment on AI during the session. Abel took the lead on strategic questions about how Berkshire will address the technology across its insurance, rail, energy and manufacturing units.

Berkshire Hathaway Energy drew particular attention. Abel said data centers already account for about 8% of peak electricity load in service areas such as Iowa, a figure near industry benchmarks of 5% to 10%. He projected the unit could increase that footprint by roughly 50% over the next five years as large cloud operators expand infrastructure for AI workloads, and he told shareholders those operators must “have to bear the full cost” so residential and commercial customers do not absorb added expenses.

Abel said Berkshire will apply clear tests before adopting AI tools: the technology must improve safety, make operations more efficient, or enhance the quality of decisions. He contrasted that selective approach with companies that have announced workforce reductions or rebranded products to emphasize AI.

Executives and investors will observe how the company balances potential demand for AI-related infrastructure with Berkshire’s longstanding capital allocation practices. Abel noted the conglomerate’s diverse businesses provide multiple points of contact with AI, from automated underwriting in insurance to operational analytics in freight rail and increased power demand from data centers.

Abel’s remarks focused on narrow, targeted deployments rather than broad software bets. He emphasized investments will be made where returns are demonstrable, and that Berkshire will not subsidize third-party AI builders through higher utility rates for its customers.