Google CEO Sundar Pichai expects his company to invest around $75 billion in capital expenditures in 2025 to boost artificial intelligence offerings.
“We expect to invest approximately $75 billion in capital expenditures in 2025,” Pichai said in a statement in Google parent Alphabet’s fourth-quarter 2024 earnings report. The figure is a 43% increase from the firm’s $32.3 billion capital expenditures in 2023.
He said the investment would “accelerate progress” in AI innovation and continue to strengthen the company’s core businesses.
Capital expenditures — sometimes referred to as “capex” — are funds used to purchase long-term physical or fixed assets in business operations.
It’s not clear exactly how much of the investment is earmarked for AI, but it’s expected that most of it will be funneled into expanding Google’s AI infrastructure. Other Big Tech firms have increased spending on AI-related projects, including Meta, which said it would spend $65 billion to expand its AI infrastructure.
AI has been one of Google’s strongest revenue streams in the last few years, with overall revenue up 12% year-over-year at $96.5 billion.
Meanwhile, Google Cloud revenue jumped 10% to $12 billion in the same time frame, something Google said had been supported by consistent growth across “core Google Cloud Platform products, AI Infrastructure and Generative AI Solutions.”
However, Alphabet’s total revenue fell short of combined analyst expectations of $96.7 billion, and its share price fell more than 7% in after-hours trading, according to Yahoo Finance.
Related: Google exposes government-backed misuse of Gemini AI
In a Feb. 4 investor call, Pichai downplayed the risks posed by new competitors, including the China-based AI model DeepSeek, which shook the market in late January.
Pichai told listeners of the call that Google’s Gemini 2.0 Flash models are “some of the most efficient models out there,” even when compared to DeepSeek’s v3 and R1 models.
On Jan. 27, DeepSeek spooked US markets with news that the model’s developers were able to produce a worthy competitor to American AI firms such as OpenAI at a fraction of the cost.
The firm claims it was able to develop its AI model on a shoestring budget of just under $6 million using less advanced hardware from semiconductor manufacturer Nvidia.
AI Eye: 9 curious things about DeepSeek R1
