The great capex divide: How Amazon and Apple are on opposite ends of the AI boom

Everyone wants to be an AI company. Not everyone wants to spend like one.

Tech companies have had to square expensive investments in artificial intelligence with the fact that much of its return on investment is so far theoretical, or at least far off.

The issue has become more acute since the arrival of China’s DeepSeek earlier this year, which unveiled a lower-cost model that used a process called distillation — essentially training on the giant frontier models bankrolled by major tech companies to produce smaller but very capable models more efficiently.

That event has crystalized two diverging camps among American tech companies: those that spend a lot investing in AI, like Amazon, and those that don’t, like Apple.

Other companies, like Alphabet, Microsoft (and, by extension, OpenAI, in whom Microsoft is a major investor), and Meta, are also in the first camp. This year, the four companies combined are set to spend more than $315 billion on capital expenditures, much of it earmarked for AI efforts. Their thinking is that even if models like DeepSeek come along and create processes by which more can be done with less, more is still in fact more. They cite Jevons Paradox, the idea that cost efficiencies will drive more demand, not less. They’re also aligning themselves with the trend toward increased performance, which uses more computationally intensive “reasoning” models. And with all the spending they’ve done, they’ll be the best positioned to reap those future rewards.

“I continue to think that investing very heavily in capex and infra is going to be a strategic advantage over time,” Meta CEO Mark Zuckerberg, who’s committed $60 billion to $65 billion to capex this year, said on the company’s earnings call in January. “It’s possible that we’ll learn otherwise at some point, but I just think it’s way too early to call that.”

“AI represents, for sure, the biggest opportunity since cloud and probably the biggest technology shift and opportunity in business since the internet,” Amazon CEO Andy Jassy said on the company’s earnings call last month. “I think that our business, our customers, and shareholders will be happy medium- to long-term that we’re pursuing the capital opportunity and the business opportunity in AI.” Amazon has committed to spending more than $100 billion in capex this year.

Then there are companies like Apple and Salesforce, whose strategy involves spending a small fraction of what their peers do on capex. Their AI ambitions are no less central to their future businesses than the others, but they’ve chosen to be more measured in spending, often renting others’ AI instead of owning, and hedging their bets with partnerships. Apple has paired with OpenAI’s ChatGPT to furnish its AI ambitions. In China, it’s working with both Baidu and Alibaba to bring AI to its iPhones.

“Innovation that drives efficiency is a good thing,” Apple CEO Tim Cook said regarding DeepSeek on the company’s latest earnings call. “From a capex point of view, we’ve always taken a very prudent and deliberate approach to our expenditure, and we continue to leverage a hybrid model, which I think continues to serve us well.”

Salesforce expects its capex to be just 2% of its revenue again this year — for comparison, some Big Tech companies are spending more like 15% to 30% of their revenue on capex — choosing to use Amazon and Google’s data centers rather than build its own. While not exactly in the same league as the others, Salesforce is still a giant tech company that fancies itself an AI company and whose leader has been explicit in regard to how Salesforce is setting itself apart from those others.

“We aren’t building huge $10 billion, $20 billion, $30 billion, $100 billion data centers. We’re not doing some of these kind of engineering efforts that may or may not have some kind of huge payoff, but is going to take down all of our cash and all of our margin for the next several years,” CEO Marc Benioff said on Salesforce’s last earnings call. He has previously described AI spending by his competitors as “excessive” and “a race to the bottom.”

“We’re augmenting our existing product line with artificial intelligence, taking advantage of these incredible investments that are being made in infrastructure by others, and we’re going to deliver the digital labor revolution,” he said.

For now, it’s uncertain which strategy will be the most successful. What we do know is that neither guarantees success.

To wit: Apple, which has spent relatively very little, is amid an AI crisis, having lagged its peers in developing a functional AI assistant. Then there’s Alphabet, which has spent a ton and continues to fork over cash for AI. Cofounder Sergey Brin recently griped that the company could reach artificial general intelligence — a term for when the AI can do tasks as well as humans — if only its human workers would work harder and show up to the office “at least every weekday.”