Shortly after his inauguration on Jan. 20, President Donald Trump hosted an event at the White House that featured some of the biggest names in the technology industry.
OpenAI CEO Sam Altman, Oracle founder Larry Ellison, and Japanese tech mogul Masayoshi Son are leading the charge for an infrastructure project called Stargate, which aims to invest $500 billion into American technology companies over the next four years.
While initial reactions to this initiative were positive, the euphoria was short-lived. Just days after the Stargate announcement, investors were focused on a new AI start-up out of China called DeepSeek.
Below, I’m going to explore some of the main talking points surrounding DeepSeek and assess if the newest AI darling poses a threat to Trump’s Stargate agenda.
What makes DeepSeek such a point of contention is that the company claims to have trained its models using older hardware compared to what AI companies in the U.S. are leveraging. For example, over the last couple of years, you’ve likely heard business leaders talking about Nvidia‘s (NASDAQ: NVDA) graphics processing units (GPU) — namely the Hopper and Blackwell architectures. These are Nvidia’s most sophisticated — and expensive — chips.
However, DeepSeek’s developers claim to have used older GPUs and less expensive infrastructure from Nvidia, primarily a cluster of H800 chips.
This is important because the team at DeepSeek is subtly implying that high-caliber AI can be developed for much less than what OpenAI and its cohorts have been spending. If that’s the case, it’s reasonable that investors would expect a major pullback on AI infrastructure spending. While I acknowledge the logic supporting the bear narrative here, let’s explore if this argument is really holding up.
Over the last couple of weeks, several big players in the AI revolution reported earnings for the fourth quarter and full year 2024.
-
Microsoft: Microsoft is a big investor in OpenAI, and over the last couple of years, the company integrated new AI-powered services throughout its ecosystem. One of the most important areas where Microsoft is leveraging AI is its cloud computing business, Azure. In the last 12 months, the company plowed more than $55 billion into capital expenditures (capex) — a lot of which is allocated toward the company’s ongoing AI efforts. Microsoft’s management appears committed to these infrastructure projects, as management told investors that the company’s capex budget for this year should be in the range of $80 billion.
-
Meta Platforms: Meta hasn’t been afraid of doubling down on its AI vision. In 2023, the company spent $28 billion in capex. Last year, Meta’s infrastructure spending rose by 40% — coming in at around $39 billion. During the company’s recent earnings call, Meta’s management said that capex spending in 2025 will be in the range of $60 to $65 billion — representing an increase of 67% year over year at the high end of the range.
-
Alphabet: Alphabet spent the last couple of years integrating AI services into its own ecosystem in an effort to diversify its business from heavy reliance on advertising and unlock new opportunities to compete more directly with Microsoft and Amazon. During the company’s fourth-quarter earnings call, investors learned that Alphabet plans to spend around $75 billion on capex in 2025, representing an increase of 43% compared to spending levels in 2024.
-
Amazon: Amazon has also spent considerable capital on AI infrastructure over the last couple of years. In particular, the company poured billions into a start-up, Anthropic, which competes directly with OpenAI. Integrating Anthropic into the cloud business, in particular, helped the company reaccelerate sales and widen profit margins in Amazon Web Services (AWS). During the company’s fourth-quarter earnings call, management alluded that Amazon could spend roughly $105 billion in AI infrastructure this year, as demand for AWS remains robust. Should Amazon spend around the indicated levels, that would represent an increase of 27% compared to capex in 2024.
