Greg Abel is stepping into one of the most difficult succession jobs in corporate America: running Berkshire Hathaway after Warren Buffett.
Since taking over as chief executive at the start of 2026, Abel has largely allowed Berkshire’s investment decisions to speak for themselves. But in a wide-ranging CNBC interview on Wednesday, he offered a clearer picture of how the company intends to navigate an economy increasingly shaped by artificial intelligence, while preserving the long-term, operating-focused philosophy that defined Buffett’s tenure.
The message is that Berkshire is not sitting out the AI boom. It is simply approaching it from a different angle.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Rather than attempting to identify the next breakthrough model or speculate on which AI company will dominate, Berkshire is positioning itself around the infrastructure that AI companies will need to operate, particularly electricity. That represents an important evolution for a company whose reputation was built on investments in insurance, railroads, energy, industrial businesses and consumer brands rather than high-growth technology.
Berkshire’s roughly $38 billion investment in Alphabet is the clearest indication that the company is willing to make a major direct bet on AI.
Abel confirmed that Berkshire began accumulating Alphabet shares last year, saying the company came to Berkshire’s attention partly because AI was already proving useful within Berkshire’s own operating businesses.
The Alphabet investment is spectacular not simply because of its size, but because it suggests Berkshire no longer views AI as a speculative technology theme that sits outside its traditional investment framework.
The more revealing part of Abel’s comments, however, was his focus on electricity.
For Berkshire, the next constraint on AI may not be chips or models but the ability to generate and deliver enough power to support the enormous data centers being built by hyperscalers. That gives Berkshire an unusual position in the AI infrastructure race. Through its energy operations, the company owns utilities serving Iowa, Nevada, and large parts of the western United States. Those businesses are directly exposed to the surge in electricity demand created by data centers.
Berkshire’s Iowa utility already obtains roughly 8% of its electricity load from data centers, giving the company an existing foothold in a market that could expand rapidly as AI developers increase computing capacity. This is where Berkshire’s traditional operating philosophy becomes particularly relevant. The company does not need to predict which AI model will win. If AI companies continue expanding data-center capacity, they will require electricity regardless of whether the eventual winner is OpenAI, Google, Anthropic or another developer.
In that sense, Berkshire is attempting to capture the infrastructure spending behind the AI boom rather than betting exclusively on the technology at its center.
But Abel’s comments also highlighted the limits of that opportunity.
Berkshire will serve new data centers only if doing so does not increase electricity costs for its existing customers. That condition is relevant because the AI infrastructure buildout is becoming a political and regulatory issue as well as a technology investment story. Data centers can require enormous amounts of electricity, forcing utilities to invest in generation, transmission, and grid infrastructure. The question therefore is who pays for those investments and whether existing households and businesses should bear part of the cost.
Berkshire’s position suggests the company sees the demand opportunity but recognizes that utilities cannot simply redirect scarce power toward hyperscalers without considering affordability and reliability for existing customers.
That could become one of the biggest constraints on the next phase of the AI boom.
The semiconductor industry has spent years worrying about whether there will be enough advanced chips to satisfy AI demand. The next bottleneck may sit further downstream: electricity generation, transmission capacity, data-center construction and regulatory approvals.
AI developers can order more GPUs, but they cannot instantly build power plants or transmission lines. That makes electricity potentially one of the most strategically valuable assets in the AI supply chain.
However, it is also a familiar business for Berkshire. Unlike a technology investor, the company can potentially participate in the expansion by owning and operating physical infrastructure that generates long-term cash flows. The strategy also fits Berkshire’s preference for businesses where management can exercise direct operational control rather than simply owning financial stakes.
Still, Berkshire enters this new era from a position of relative underperformance. Its shares have barely moved this year and have lagged the S&P 500 by more than 10 percentage points as the market’s AI-driven rally has rewarded technology and growth stocks. But some analysts believe that it creates pressure for Abel, although it is unlikely to change Berkshire’s fundamental investment philosophy.
The challenge is that Berkshire’s enormous cash holdings and concentration in traditional businesses can look unattractive when investors are aggressively rewarding companies exposed to AI.
Abel’s response appears to be less about transforming Berkshire into a technology company and more about finding where the AI boom intersects with businesses Berkshire already understands.
That approach extends beyond technology and energy. Abel said consumers remain under pressure from inflation and high mortgage rates, while the housing market faces a “bumpy road” without an immediate recovery.
Yet Berkshire recently bought homebuilder Taylor Morrison, indicating that it is willing to invest in industries experiencing near-term weakness when it believes the long-term economics remain attractive.
Abel expects Taylor Morrison to become a “very strong asset” over five to 10 years, noting that the underlying demand for homeownership will remain even if high housing costs are preventing many Americans from buying homes today.
That investment offers another clue about how Abel intends to run Berkshire.
The company does not necessarily need the economy to be strong everywhere at the same time. It can deploy capital into sectors where temporary weakness creates attractive long-term opportunities, while allowing its operating companies to benefit when conditions improve. The result is a Berkshire that is gradually becoming more exposed to the forces driving the modern economy without abandoning the principles that made it successful.
Alphabet gives Berkshire direct exposure to AI. Its utilities provide exposure to the electricity required to run AI infrastructure. Taylor Morrison provides exposure to a long-term housing shortage. And its traditional businesses continue generating the cash needed to fund those investments.
That may ultimately prove to be Abel’s biggest test.
Buffett built Berkshire around patience, capital allocation and the ability to look beyond short-term market enthusiasm. Abel now has to apply those principles to an economy in which technological change is occurring at extraordinary speed.
His answer so far is not to chase the AI trade.
It is to identify what the AI boom will need next, invest in those physical and economic bottlenecks, and wait for the demand to translate into durable cash flows.