Meanwhile, its Gemini app has grown to 950 million monthly active users.

That puts its AI chat app within striking distance of OpenAI’s ChatGPT which hit one billion monthly active users in June, according to reports.

Alphabet spent nearly US$45b in the quarter, “the vast majority” of which went towards AI investments, CFO Anat Ashkenazi said during a call with analysts.

The company also increased its capital expenditure estimate for the full year to as much as US$205b, a jump from US$190b last quarter.

The increase is “primarily due to an acceleration in the delivery of capacity to meet growing demand”, Ashkenazi said. “We continue to expect our capex [capital expenditure] to increase significantly in 2027, and will provide more details at a later date.”

The company’s share price fluctuated after hours and dropped by more than 4%.

AI race

Major tech companies in the US have been spending billions of dollars to build out AI infrastructure and develop increasingly more advanced AI models.

In June, Google parent company Alphabet announced it would raise up to US$80b in stock to fund its AI efforts, with Warren Buffett’s Berkshire Hathaway committing US$10b.

The spending is part of a larger race across the tech sector to stay competitive, amid soaring demand for computing power.

Amazon, Microsoft, Alphabet and Meta are collectively on track to pour roughly US$700b into AI data centres, chips and computing infrastructure this year.

Amid competition for customers, AI developers are also designing lower-cost versions of their own models.

On Tuesday (local time), Google launched three new models under the name “Flash”, which the company described as cheaper, faster and more reliable.

Cheaper and faster models coming from developers in China – particularly from DeepSeek and Moonshot – have ruffled feathers and raised concerns about intellectual property theft in the US.

This week, White House technology official Michael Kratsios accused Moonshot of creating its new Kimi K3 model by distilling Anthropic’s powerful new model, known as Fable.

Distillation is a technique in which a smaller AI model is trained to mimic the outputs of a larger, more capable one.

– AFP