Google commits $190B to AI infrastructure as it unveils new TPUs

Alphabet CEO Sundar Pichai announced at Google I/O 2026 that the company expects to spend approximately $190 billion in capital expenditures this year, representing a six-fold increase from the $31 billion spent annually in 2022. The massive infrastructure investment will fund AI data centers, model training, and the company's own custom Tensor Processing Units, signaling a strategic push toward self-reliance in semiconductor supply. Meanwhile, Nvidia reported second-quarter revenue guidance of $91 billion, plus or minus 2 percent, beating Wall Street estimates of $86.84 billion, and announced an $80 billion share repurchase program alongside a dividend increase from one cent to 25 cents per share. Google unveiled its eighth-generation custom silicon featuring a dual-chip architecture: the TPU 8t optimized for large-scale model pretraining delivering nearly three times the computing power of the previous generation, and the TPU 8i designed exclusively for inference to handle live user queries with reduced latency. Pichai emphasized that training can now scale across more than 1 million TPUs globally, enabling developers to train larger models in weeks rather than months. The company reported that over 8.5 million developers build applications using Google AI models monthly, with APIs processing roughly 19 billion tokens per minute, while the Gemini app has crossed 900 million monthly active users, more than doubling its user base since last year. The announcements come as U.S. tech giants including Alphabet, Amazon, and Microsoft are expected to spend more than $700 billion on AI infrastructure this year, up from approximately $400 billion in 2025. While these companies remain heavily reliant on Nvidia processors, they are simultaneously investing in custom chips targeted primarily at inference, the process by which AI responds to queries, which represents a much larger market than training. Nvidia, widely considered a barometer for AI market health, faces increasing competition from Big Tech and chip rivals like Intel and AMD, prompting the Santa Clara company to recently unveil new products built on Groq technology to defend its market position.