Sundar Pichai flags ‘intense’ 2026 as Google plans to ramp spending to meet AI pressure

Google

Google’s top AI infrastructure executive has told employees that the company now needs to double its compute capacity every six months to meet soaring demand for AI services. According to a report by CNBC, at an all-hands on November 6, Amin Vahdat, vice president, Google Cloud, said the next target is a 1,000-times jump in capability over the next four to five years. He stressed that the race to build AI infrastructure is both the most expensive and the most decisive battleground.

Alphabet CEO Sundar Pichai and CFO Anat Ashkenazi were present as Vahdat outlined why Google is accelerating spending. The company recently raised its capital expenditure guidance to roughly $91 billion to $93 billion for the year, with an even larger increase expected in 2026. Rival hyperscalers Microsoft, Amazon and Meta have all lifted their own capex forecasts, with the four together set to spend over $380 billion.

Pichai addresses bubble fears and cloud momentum

With fears of an AI bubble resurfacing, Pichai admitted concerns are valid but said the bigger risk is underinvesting. He pointed to Google Cloud’s 34 percent annual revenue jump to more than $15 billion, adding that numbers could have been stronger if more compute had been available.

Pichai said 2026 will be “intense”, driven by AI competition and capacity constraints. He noted that tools like Veo, Google’s upgraded video generator, could have reached more users if infrastructure limits hadn’t held the company back.

Custom silicon and efficiency at the core

As per the report, Vahdat said the answer is not simply to outspend competitors but to deliver infrastructure that is more reliable, efficient and scalable. Google is leaning heavily on custom silicon, including its newly launched seventh-generation TPU, Ironwood, which is nearly 30 times more power-efficient than the first-generation Cloud TPU. The company also expects DeepMind research to give it an edge in predicting future model requirements.

The challenge, however, is steep. Google must deliver exponentially more compute, storage and networking at roughly the same cost and energy usage. Vahdat acknowledged that achieving this will demand deep collaboration across teams.

Ashkenazi, meanwhile, said Google is focused on maintaining healthy cash flow while expanding cloud opportunities. Despite market jitters, leadership maintains that the long-term AI opportunity is simply too large to ignore.