Berkshire Hathaway CEO Greg Abel spoke with CNBC’s Becky Quick on Wednesday about the company’s big investment in the conglomerate’s parent company, Google. alphabetcalled the tech giant a “key player” in artificial intelligence.
Abel said Berkshire clearly understands the impact of AI across its corporate portfolio, and Google’s position on emerging technologies is strong.
“We started getting a lot more information internally about how we were using AI and what benefits it was bringing us. So there was a lot of interest and we recognized Google as an important player,” he told Quick.
Berkshire acquired $10 billion in Google stock at a 6.5% discount 15 months ago, Abel said.
“They hadn’t decided on a size, but they recommended we look at $10 billion, and Warren and I talked about the size. We talked about the size of the discount, and I recommended a 6.5% discount, and we were comfortable with that,” Abel said. “Then the deal was finally completed.”
Berkshire added $17 billion in Alphabet stock in the second quarter, making it the third-largest holding in Berkshire Hathaway’s stock portfolio, according to the company’s second-quarter portfolio snapshot filed with the Securities and Exchange Commission in August.
GOOGL From the beginning of the year to today
As of the company’s last filing, Berkshire owned about 106 million shares of Alphabet’s Class A and Class C shares, currently worth about $36.6 billion. The Alphabet acquisition was the biggest addition to Berkshire’s portfolio in the quarter, but Berkshire also bet big on Delta Air Lines, increasing its position by 44%, or about $1.6 billion.
Google is one of five so-called hyperscalers making huge capital investments to expand computing power to run AI. microsoft, meta, Amazon and oracle. According to investment bank Goldman Sachs, global hyperscaler capital spending is estimated at about $1 trillion in 2026, but estimates vary widely.
“Our results show that the commonly cited hyperscaler capex forecast of $794 billion likely underestimates total global AI capex by about $200 billion. At the same time, the $794 billion figure likely overestimates U.S. AI investment by $200 billion,” Goldman Sachs researchers said in an August analysis.
The question for investors is how quickly these large investments will yield returns.
AI computing workloads typically fall into two categories: large datasets and algorithm training based on inference or query responses.
Goldman Sachs researchers wrote in May that “a longer training-intensive phase lengthens ROI timelines as capital investments and R&D continue to be deployed prior to broader monetization.”
