Google's parent company Alphabet has raised its 2026 capital expenditure forecast to between US$195bn and US$205bn, spooking Wall Street even as Google Cloud revenue surged and CEO Sundar Pichai revealed a strategic pivot of compute towards training the next-generation Gemini 4 model.
Cloud Growth Beats Expectations
Alphabet's latest quarterly report delivered strong results for Google Cloud, with second-quarter revenue up 82% to US$24.8bn, well ahead of the average analyst expectation of a 64% increase, according to data compiled by LSEG.
Total revenue for the quarter also beat forecasts, coming in at US$119.8bn against a consensus estimate of US$116.9bn. Advertising revenue, meanwhile, landed closer to expectations at US$81.6bn versus an estimated US$81.1bn.
Sinking Into the Negatives
Behind the growth story sits a costlier reality. Anat Ashkenazi, Google's Chief Financial Officer, now expects 2026 capex to land between US$195bn and US$205bn, up from a prior range of US$180bn-US$190bn, which itself was a US$5bn increase on the quarter before.
The hike unsettled investors, sending Alphabet's shares down roughly 3% in extended trading. For the first time ever, Alphabet also reported negative free cash flow, burning US$5.9bn as quarterly spending hit US$44.9bn, an outlay that exceeded operating cash flow entirely.
"After a negative cash flow quarter, the new raise in capex does not sit well for Alphabet. The market's most reliable cash generators are now spending more than they bring in. As long as revenue keeps accelerating, investors will tolerate it, but capital has a real cost again and the room for error is shrinking every quarter."— THOMAS MONTEIRO, SENIOR ANALYST, INVESTING.COM
A High-Stakes Bet on Gemini 4
By funnelling resources into next-generation frontier training rather than patching a delayed release, Google is betting on a generational leap over a defensive fix. But skipping or delaying a model step carries its own risk, since developer habits tend to be sticky once workflows are built around a given model.
For Google to hold its ground in the AI race, Gemini 4 needs to land before coders and enterprises standardise their workflows around the rival models that got there first.