Pichai pushes back on claims Google is losing ground in AI race

Alphabet nearly doubled AI infrastructure spending, with Q2 results topping estimates, but investors remained focused on Gemini 3.5 Pro delays in AI coding/agent tasks. Capex guidance was raised by $15B to $195B–$205B as cloud growth surged 82% (vs. 64% average estimate), yet shares fell over 3% after hours and are down ~9% since late April amid concerns about AI momentum.

Analysis

The market should separate two businesses inside Alphabet: the durable cash engine and the increasingly expensive option on frontier AI. The capex step-up is positive for the AI supply chain, but it also signals that Google may need to spend more just to defend share, which lowers incremental return on capital versus peers that are still monetizing AI with less visible capital intensity. That is a subtle negative for the multiple: if investors conclude AI is forcing a permanent higher depreciation/capex load without a proportionate revenue step-up, free cash flow deserves a lower terminal yield.

Near term, the main risk is not a collapse in demand but a credibility gap. If model cadence continues to slip, the stock can remain under pressure for 1-3 months because investors will extrapolate execution risk into product adoption and cloud win rates. The main falsifier is a clean sequence of releases plus evidence that AI features convert into measurable cloud margin expansion or ad conversion uplift; absent that, the capex increase reads as defensive spending rather than share-taking investment.

The contrarian view is that consensus may be overweighting the frontier-model narrative and underweighting distribution. Most monetizable AI workloads are inference-heavy and cost-sensitive, where cheaper “workhorse” models can matter more than headline benchmark wins. If Google can monetize that layer while keeping search and cloud intact, the selloff is likely more of a sentiment reset than a fundamental break; if not, the overhang becomes structural over 6-18 months as capital returns get crowded out by the AI arms race.

Key Decisions for Investors

• Short GOOGL / long MSFT for 1-3 months: expresses the view that AI credibility and developer mindshare are accruing faster to the better-executing platform, while Alphabet's higher capex intensity should compress its FCF multiple. Cover if Google shows two consecutive model releases on schedule and cloud growth stays above management's implied run-rate.
• Buy NVDA on weakness, but size it as a supply-chain hedge rather than a pure Google winner trade: Alphabet's spend still supports accelerator demand broadly, but Google’s use of custom silicon means the upside is less direct than the headline capex number suggests. Falsify if Alphabet explicitly shifts more of its build toward in-house TPU substitution and external GPU demand slows.
• Use the post-earnings drawdown to add to a small GOOGL position only if the stock keeps selling despite stable cloud margins: this is a conditional mean-reversion setup, not a conviction long. The key catalyst is next quarter's commentary on AI revenue per dollar of capex and the cadence of Gemini releases.
• Watch for relative underperformance in AI-infrastructure proxies versus Alphabet over the next 2-6 weeks: if GOOGL stabilizes while semis/rack-and-networking names lag, the market is signaling that the spend is defensive and internally absorbed, not a broad-based capex boom.
• Set an alert on any further delay in Gemini 4 or evidence that partner testing does not convert to broad deployment: that would extend the multiple compression window and make a deeper de-rating in GOOGL more likely over the next 6-12 months.