First time in history! Google is burning through AI computing power, causing a “negative cash flow,” and while Q2 revenue looks impressive, the stock price still fell 4.5%.

Under the high-pressure AI arms race, tech giants also have no choice but to endure severe shocks to their financial structure. According to Ars Technica, Google’s latest 2026 Q2 financial report shows that while total revenue soared to $119.8 billion and beat market expectations, its quarterly capital expenditures for building AI data centers surged to $44.9 billion, leading the company to post its first “negative free cash flow” since going public (-$5.8 billion). The market is worried about the payback speed of high-volume AI investment, dragging the stock price down by about 4.5%.
（Background: Google’s financials massively beat expectations, but capital expenditure doubled to as high as $205 billion, and the stock received no goodwill）
（Background addition: Google launched three new models, including Gemini 3.6 Flash! 3.5 Flash-Lite is blazing fast, and Gemini 4 pre-training has already started）

Table of contents

Toggle

• Q2 revenue $119.8 billion, cloud growth strong
• AI capital expenditures drag cash flow, full-year forecast raised to $205 billion
• Facing strong rivals like GPT-5.6 lurking, stock plunges 4.5%

Against the backdrop of the global generative AI boom, tech giants are burning money on an unprecedented scale. Today in Taipei time, Google released its 2026 second-quarter financial report. Although the report delivers an impressive performance on revenue figures, it also brutally reveals the astonishing pace of capital consumption behind the AI infrastructure race.

Q2 revenue $119.8 billion, cloud growth strong

From a fundamental perspective, Google’s operations remain very robust. The financial report data shows total revenue for the quarter reached $119.8 billion, ahead of analysts’ expectations. Among them, search ads remain the biggest cash cow, contributing $63.3 billion; and thanks to strong demand for AI services, Google Cloud (cloud business) revenue reached $24.8 billion, up sharply 23.8% from the prior quarter.

In addition, the subscriptions, platform, and devices segment contributed $12.9 billion; YouTube ads grew by more than 12% to $11.1 billion due to factors such as longer ad time. After excluding non-cash earnings, Google’s operating cash flow totaled $39.1 billion, up 40% year over year.

AI capital expenditures drag cash flow, full-year raised to $205 billion

However, this seemingly impressive report contains a warning that shocked Wall Street. To secure its leading position in AI, Google’s capital expenditures (Capex) mainly used for AI data centers this quarter came to $44.9 billion—far exceeding its operating cash flow. This directly resulted in Google recording negative free cash flow (Free Cash Flow) of -$5.8 billion in the quarter—a historic first for the company since going public.

What has drawn even more attention from the market is that Google’s leadership admitted that such high-level AI investment has become “the new normal,” and next year’s spending is expected to keep rising. The company raised its 2026 full-year capital expenditure forecast further from the previous $8T to $190 billion range, to as high as $205 billion. Compared with 2022’s mere $22.0 billion in capital expenditures, the figure surged nearly 9x over just four years.

Facing strong rivals like GPT-5.6 lurking, stock plunges 4.5%

Although Google currently still holds a massive cash reserve of more than $100 billion and fundamentals remain solid, investors have clearly started to doubt the AI track of “high costs with low immediate profits.” Forecasts suggest AI capital expenditures across the entire tech industry this year will exceed $700 billion. With near-term returns still uncertain, after the earnings release, Google’s stock price immediately fell by about 4.5%, and has continued to trend downward.

Beyond financial pressure, Google also faces attacks on the technical product front from both inside and outside. Recently, there have been reports that the release of the flagship model Gemini 3.5 Pro has been forced to be delayed, with it currently only available in limited-scope testing. It also faces direct competition from strong rivals such as OpenAI’s GPT-5.6 and Anthropic’s Claude Mythos. As many top AI researchers have recently left one after another, the next few months will undoubtedly be the most critical battle for determining whether Google’s AI dominance can continue. The good news is that, according to rumors, its latest in-house AI chips (Tensor 8i and 8t) have achieved significant improvements in data center efficiency performance, which may provide some breathing room for future compute cost.