Google's Q2 earnings report will shape the AI market outlook.

Article by: Tide Research

At 4:00 AM Beijing time on July 23, Alphabet will release its Q2 2026 earnings report, followed by an earnings call at 4:30 AM. This is the first earnings report among the "Magnificent Seven," with Microsoft, Meta, Amazon, and NVIDIA to follow. The options market has already priced in this event: implied volatility of 5.3% implies a single-day price swing of over $200 billion for a company with a $4.5 trillion market capitalization.

The market is trading it like a corporate earnings report, but in reality, it’s a midterm exam for the entire AI capital expenditure cycle.

Answer correctly, and the upcoming earnings seasons for the following giants will go smoothly; answer poorly, and no one—from Nasdaq to A-share optical modules—will be spared.

On July 16, Bloomberg, citing ten current and former Google employees, reported that Gemini 3.5 Pro is months behind schedule due to its programming capabilities falling short of internal standards. On the same day, Alphabet's stock closed down 4.4%, erasing approximately $200 billion in market value.

The reason to sell in the market has nothing to do with revenue or profit—it depends solely on one shipping schedule.

Pichai personally promised at the I/O conference in May that it would be released in June; after June passed, Google only offered the vague statement, "Currently testing with partners." Bloomberg’s report filled in the remaining details: the training data was updated once at the end of June, with a focus on programming, but the results disappointed internal teams; engineers were demoralized, and some researchers have moved to competitors.

Programming is currently the most valuable battleground for commercializing large models, as it directly determines whether AI agents can independently complete multi-step tasks and ultimately decides which platform enterprise customers will pay. OpenAI and Anthropic are releasing new models weekly, while Gemini 3.5 Pro lags behind by a week—each passing week deepens enterprise developers’ commitment to competing platforms.

This downturn has revealed a new contract for AI trading in 2026: the market will tolerate you burning nearly $200 billion in a year, provided the model arrives on time and is powerful enough upon arrival. Once the schedule slips, capital expenditures revert from a "moat" back to a "bill."

Tonight’s earnings report is Google’s first accountability under this new agreement.

Market consensus is clear: revenue of approximately $116.8 billion, up about 21% year-over-year; earnings per share of $2.86 to $2.89, up approximately 24% year-over-year. Google has exceeded expectations for the past four quarters; simply beating estimates no longer drives the stock price—the real variables lie in three key indicators.

Cloud growth and conversion of backlog orders.

In the first quarter, Google Cloud's revenue grew 63% year-over-year, significantly outpacing Azure's 40% and AWS's 28%, with backlog orders nearly doubling to $462 billion.

CFO Ashkenazi left a key statement last quarter: if capacity had kept pace with demand, cloud revenue could have been higher. This statement distills the issue to its core—tonight, we’ll verify the speed of capacity ramp-up and determine how much of the $462 billion is progressing at the management’s promised pace (approximately half within 24 months). If growth holds above 60%, the story of AI infrastructure continues; if it slows noticeably, the market will immediately revalue all cloud providers.

Where the money went and the hint toward 2027.

Google’s 2026 capital expenditure guidance of $180 billion to $190 billion is more than five times its 2023 spend of $32.3 billion, and management has signaled a “significant increase” by 2027. The cost is reflected on the cash flow statement: free cash flow declined nearly 47% year-over-year in the first quarter, and another $80 billion in stock was issued in June to fund AI expansion. Tonight’s earnings call may see any specific figures regarding 2027 spending levels have a greater impact on the stock price than the Q2 results themselves. An increase in spending paired with strong cloud growth is aggressive; an increase in spending paired with slowing cloud growth is catastrophic.

TPU has transitioned from a cost advantage to a second revenue stream.

This is the most speculative undercurrent in this earnings report. Google has begun selling TPU computing power to external customers; a data center leasing company backed by Blackstone plans to rent out TPUs to AI labs starting in 2027; Anthropic has signed a letter of intent for a million-unit order of Ironwood chips; and Meta is reportedly considering deploying TPUs in its own data centers by 2027. Management previously indicated that hardware sales would begin contributing revenue in the second half of 2026, with the bulk expected in 2027. Even a single confirmation tonight about the progress of TPU external sales could be enough to reprice the "Google chain."

The reach of Google's earnings report has long extended beyond U.S. stocks.

At the U.S. stock market level, as the first major company to report, Google’s earnings call will set the tone for the entire earnings season. Cloud growth rates and capital expenditure guidance will be directly incorporated into the expected models for Microsoft and Meta; any progress on TPU will serve as a direct stress test for NVIDIA. The memory of last November, when rumors emerged that Meta was negotiating to purchase TPU, causing NVIDIA’s stock to plunge in a single day and forcing the company to publicly declare that its technology still led the industry by a generation, remains fresh.

At the A-share level, "Google Chain" is already one of the most weighted narratives in the computing power rally.

The deployment of TPU clusters (a single cluster with 9,216 liquid-cooled chips interconnected via OCS all-optical switching) results in significantly higher per-unit demand for optical modules, optical switching, liquid cooling, and high-speed PCBs compared to traditional architectures. Zhongji旭创 and Xinyisheng have risen to the top three weightings on the ChiNext Board, with order visibility extending to 2027; SinoMicroelectronics is the exclusive foundry for Google’s OCS MEMS micromirror chips. Tonight’s capital expenditure guidance will serve as the benchmark for these companies’ order levels over the next six quarters. An upward revision in spending will automatically resolve valuation disputes around optical modules; if spending falls short of expectations or cloud growth slows, the crowded AI compute trade could face its most severe sell-off of the year.

Cloud exceeded expectations, with spending节奏 under control. Google Cloud maintained growth above 60%, provided guidance for 2027 spending without causing alarm, and confirmed the commercialization of TPU sales. The stock is challenging its previous high of $408, as AI trading sees a broad recovery, with China’s A-share computing chain following suit. This is the bull case—and the path implied by the 43 analysts who have issued buy ratings with an average target price of $428.

Beat earnings, but concerns over uncontrolled spending intensified. Revenue and profit both exceeded expectations, yet 2027 spending projections appear overly aggressive; free cash flow continued to deteriorate, and Gemini’s new timeline remains absent. The stock rose sharply before retracing, as market sentiment shifted toward scrutiny of “revenue growth without profit growth,” passing pressure on to Microsoft and Meta a week later.

Cloud slowdown. A drop in growth below 55%, or underperformance in backlog conversion. This is a tail risk that, if realized, would impact all cloud providers and the entire computing supply chain, including NVIDIA.

Based on the current position structure, the third scenario has the lowest probability, while the second is undervalued. The market is overly focused on whether earnings beat expectations; the true odds lie in the Q&A section of the earnings call.

At 4:30 AM, the conference call begins, and the analyst’s first question will most likely not relate to the Q2 numbers. Everyone wants to know the same thing: When exactly is the new launch date for Gemini 3.5 Pro?