According to Greg Abel, artificial intelligence only makes sense where it demonstrably improves productivity, margins or decision quality - not as a mandatory "AI-first" sticker in investor presentations.
To pour capital into AI projects just because others are doing it , he says, is to ignore the cost, return and actual impact on the business - exactly the type of behaviour that has ended in painful corrections to tech bubbles in the past.
"We're going to create technology rather than buy it"
Abel said Berkshire $BRK-B is shifting from the role of "technology buyer" to that of internal builder. In practice, that means a preference for in-house solutions - such as predictive maintenance and operations optimization on BNSF railroads or advanced analytics in the insurance industry - rather than large, expensive external AI programs or chasing "hot" AI stocks just because they're trendy.
A key phrase he repeated in various iterations, "We're not going to do AI because it's trendy." This essentially draws a line between technology as a useful tool (improving efficiency, safety, decision making) and technology as an expensive logo for a presentation that is just meant to signal "we are AI-first."
Record cash and discipline as a counterweight to AI "capex mania"
Abel's warning comes as tech giants plan to spend hundreds of billions of dollars on AI infrastructure and datacenters this year. Estimates reckon that Alphabet, Amazon, Meta and Microsoft combined will increase capital spending from roughly $410bn in 2025 to around $700bn in 2026.
On the other side of the spectrum, Berkshire is sitting on a record $397 billion in cash and short-term securities, and Abel reiterated that "capital discipline is the rule." The firm, he said, will "act decisively when the dislocation occurs and the price is right" - that is, when the hype subsides and a truly attractive risk-reward ratio emerges.
In other words, while part of the market is trying to outpace the competition with the speed and size of AI capex, Berkshire is consciously choosing the role of patient observer with ready ammunition. Abel is implicitly warning that the current environment is not ideal for capital rationality - money is too cheap for "AI projects" but too expensive for mistakes.
AI as a tool, not a label: what Abel is saying to investors
Abel's statements about AI have two levels - the internal (how Berkshire is using AI) and the external (what he's saying to tech investors):
Inside Berkshire: AI, he says, pays off where it improves efficiency, safety or decision-making - for example, at insurers Geico or BNSF, where it's predictive maintenance, performance planning and fraud detection or deepfake risk. It's not about "big stories," it's about tangible savings and better underwriting.
For non-Berkshire investors: the main message is that chasing AI companies in a presentation without proof of real returns is dangerous. Abel "draws a line between useful tech tools and expensive hype" - and he's saying that much of today's AI investing is more fashion than fundamentals.
This caution doesn't mean Berkshire doubts AI - on the contrary, Abel openly says that "AI is impacting every aspect of Berkshire" and that the technology will be key to the continued profitability of Geico, BNSF and the energy businesses. The difference is that AI must first demonstrate a tangible benefit, only then will it receive capital - not the other way around.
Rejecting the "momentum play" in AI stocks
For technology investors, Abel's blasting of the "momentum approach" is most important. He said at the meeting that Berkshire doesn't consider tech spending or tech stocks a game where you have to quickly jump on a fashion wave and hope it gets overpaid by even more enthusiastic buyers.
In contrast to recent years, when the "Magnificent Seven" attracted huge amounts of capital purely on the "AI-first" story of the future, Abel is telling shareholders that at Berkshire, the decision is still the same: the ability of technology to generate sustainable cash flow, not the number of times the word "AI" appears in the earnings call transcript.
His stance is a particularly sharp contrast to the aggressive tone of many tech CEOs who frame AI as "the new electricity" and imply that those who don't invest fully today won't exist tomorrow. Instead, Abel argues that the biggest risk today is not to ignore AI, but to blindly adopt fashionable investment theses without regard to cost or return.
What Abel's warning implies for technology investors
Three practical lessons can be derived from Abel's statements for anyone with a portfolio overloaded with AI names:
AI is not a self-justifying investment: the fact that a company is talking about AI or increasing AI capex says nothing about future returns. What matters is whether the technology improves margins, increases productivity or opens up new, monetisable markets.
Price and timing still matter: with nearly $400 billion in cash, Berkshire is waiting for a "dislocation" - a period when the hype cools and quality assets start selling at a discount, not a premium. If the world's biggest "value machine" doesn't want to buy AI stories at current prices, that in itself is a signal.
Technology vs. hype: investors should be able to distinguish between tangible use-cases that generate measurable cash flow and corporate "AI statements" in presentations. Abel says bluntly: technology should be an add-on to the business, not a fashion accessory for shareholders.