Alphabet’s AI capex is not crowding out earnings yet, and the market is still underpricing Search-to-Cloud operating leverage

The print changes the debate from “can Alphabet spend enough to defend AI relevance?” to “how much of the AI demand curve is already being monetized through Search, Cloud, and subscriptions before the depreciation wave fully arrives?” What was priced in looked like a high-expectations mega-cap quarter: Street revenue at $111,332.4 million and EPS at $2.57 already assumed healthy growth, while investors were braced for another step-up in AI infrastructure spending. What actually surprised was cleaner: revenue landed at $113,828.0 million, a +2.2% surprise, and EPS landed at $2.82, a +9.7% surprise, while management simultaneously disclosed 2026 capex of $175 billion to $185 billion. That combination matters because it undermines the simple bearish read that AI capex is immediately dilutive. Alphabet is spending into demand, not ahead of a void, and the demand evidence is now too large to treat as narrative: Search revenue grew 17%, Cloud revenue grew 48%, and backlog grew 55% quarter over quarter to $240 billion.

The revenue trajectory supports that thesis because Q4 was not an isolated beat against a low bar. In the historical series, reported revenue moved from $96,469.0 million in Q4 FY2024 to $113,896.0 million in Q4 FY2025, with revenue YoY improving from +11.8% to +18.1%, while gross margin expanded from 57.9% to 59.8%. The sequential pattern also matters: Q4 FY2025 revenue increased +11.3% QoQ after Q3 FY2025 had already grown +6.1% QoQ, so the holiday quarter was not just seasonal strength against weak prior-quarter demand. The next reported quarter in the data pack, Q1 FY2026, shows revenue of $109,896.0 million, revenue YoY of +21.8%, and gross margin of 62.4%, which is the number that challenges the consensus fear most directly. If the capex cycle were already overwhelming monetization, gross margin would not be moving from 59.8% in Q4 FY2025 to 62.4% in Q1 FY2026 in the provided history.

The margin argument is sharper when separated from the expense line, because the company is clearly paying up for the AI cycle. Anat Ashkenazi said total cost of revenue was $45.8 billion, up 13%, with other cost of revenues at $29.2 billion, up 13%, driven by depreciation tied to technical infrastructure, content acquisition costs for YouTube, and technical infrastructure operations costs. She also said total operating expenses were up 29% to $32.1 billion, with R&D expense up 42% driven by compensation and depreciation. Those are not small numbers, and they explain why the stock can still trade with capex anxiety after an EPS beat. But the variant perception is that the expense pressure is being matched by revenue pools large enough to fund it: operating income increased 16% to $35.9 billion and operating margin was 31.6%, while net income increased 30% to $34.5 billion and EPS increased 31% to $2.82, per Ashkenazi. The key read is not that costs are benign; it is that AI-related depreciation and R&D acceleration have not broken the earnings algorithm.

That earnings algorithm is still anchored in Search, and the print gives a direct answer to the market’s fear that AI interfaces would cannibalize the core faster than Alphabet could adapt. Sundar Pichai’s wording matters because he tied the acceleration explicitly to the largest profit pool, saying, “This quarter, Search continued to accelerate with revenues growing 17%, YouTube's annual revenues surpassed $60 billion across ads and subscriptions.” On the segment detail, Google Search and other advertising revenues increased 17% to $63.1 billion, with the largest contribution from retail, while Google services revenues increased 14% to $95.9 billion and Google services operating income increased 22% to $40.1 billion at a 41.9% operating margin. That is the cleanest rebuttal to a bear case centered on Search disruption: the business most exposed to AI interface risk is also the business contributing the largest disclosed revenue base, $63.1 billion, and it is growing at 17%, not decelerating into the transition.

The second leg of the thesis is that Cloud has moved from optional upside to a capex justification with numbers large enough to matter. Pichai said, “Cloud significantly accelerated with revenues growing 48% now on an annual run rate of over $70 billion.” He then gave the backlog number that should anchor forward estimates: backlog grew 55% quarter over quarter to $240 billion, driven by demand for AI products. A $240 billion backlog does not guarantee near-term revenue recognition, but it does change the burden of proof around the $175 billion to $185 billion 2026 capex range. The capex plan is not being presented against vague AI enthusiasm; it is being paired with 48% Cloud revenue growth and 55% backlog growth. The market may be mispricing the capex line as a one-way hit to free cash flow, when the operating data says Alphabet is converting AI infrastructure into booked demand at a scale that few peers can replicate.

The cash flow bridge is where bulls and bears can both find ammunition, but the better interpretation is that Alphabet has room to spend without losing capital flexibility. Ashkenazi said operating cash flow reached $52.4 billion in the fourth quarter and $160.5 billion for the full year, translating into free cash flow of $24.6 billion in the fourth quarter and $73.3 billion for the full year. Against that, capex was $27.9 billion for the fourth quarter and $91.4 billion for the full year, and the 2026 guide is $175 billion to $185 billion. The conflict is real: free cash flow will be pressured if capex nearly doubles on the provided full-year numbers, and depreciation already increased by nearly billion dollars or 38% from $15.3 billion in 2024 to $21.1 billion in 2025. But the Street should not stop at “capex up.” The relevant question is whether revenue growth, Services margin, and Cloud backlog can carry the asset build. This quarter’s answer is yes, with consolidated revenue up 17% on the company’s call basis to $113.8 billion and operating income up 16% to $35.9 billion.

The call delivery complicates the read, because management’s prepared message was more constructive on guidance than the Q&A sounded. The tone history shows Q4 FY2025 sentiment at 0.52, unchanged from Q3 FY2025 at 0.52, while guidance_tone rose to 0.65 from 0.52. That is the highest guidance_tone in the table, above Q1 FY2025 at 0.51 and Q2 FY2025 at 0.49. Yet tone_confidence fell to 0.36 from 0.43 in Q3 FY2025, and qa_evasiveness jumped to 32.0 from 2.2, while uncertainty moved to 31.9 from 30.4. The numbers say management wanted investors focused on the demand and capex commitment in prepared remarks, but was less clean in Q&A delivery. That split matters for positioning: the print is investable because the financials and backlog are concrete, but multiple expansion may wait until the Q&A tone catches up with the prepared guidance stance.

The later tone data in the pack reinforces the same point rather than invalidating it. Q1 FY2026 sentiment fell to 0.47 from Q4 FY2025 at 0.52, guidance_tone dropped to 0.28 from 0.65, and qa_sentiment moved to 0.25 from 0.51, even as prepared_sentiment rose to 0.68 from 0.01 and qa_evasiveness fell to -1.5 from 32.0. That is conflicting evidence: Q1 FY2026 shows a less positive overall and guidance tone, but also a far less evasive Q&A. For the stock, that means the next leg depends less on management sounding excited and more on whether the hard numbers continue to validate the spending. The Q4 call created the thesis with 17% Search growth, 48% Cloud growth, $240 billion backlog, and $175 billion to $185 billion 2026 capex; the next call needs to prove those dollars are turning into revenue and gross margin rather than only depreciation.

The supply chain read-through is unusually direct because Alphabet’s capex guide is now a demand signal for named AI infrastructure suppliers, not just an internal budget line. Inventec Corporation is tied to Google TPU AI-server L6 motherboards and reported new wins for L10 system and L11 rack-level integration shipped from a new Texas plant, so the $175 billion to $185 billion 2026 capex range and $27.9 billion Q4 capex support visibility into rack-level AI server demand. TSMC has exposure through 3nm/5nm custom AI TPU fabrication, while Broadcom is the TPU ASIC silicon implementation partner at approximately $8B/yr. Arm Holdings is exposed through ARM architecture for custom cores, and Nuvoton Technology through BMC baseboard-management chips and OpenTitan-class security silicon designed into Google TPU v6/v7 AI server platforms and traditional servers. The magnitude to carry through supply-chain models is the capex step from $91.4 billion in 2025 to $175 billion to $185 billion in 2026, paired with the $240 billion backlog and 55% quarter-over-quarter backlog growth that justify the build.

The peer comparison also argues that Alphabet is not just another AI spender with a lower-margin ad business attached. In the peer table, GOOGL’s latest reported quarter shows revenue of $109,896.0 million, gross margin of 62.4%, and revenue YoY of +21.8%. That growth is above AMZN at +16.6% on $181,519.0 million of revenue and AAPL at +16.6% on $111,184.0 million, while GOOGL’s 62.4% gross margin is below MSFT at 67.6%, NVDA at 74.9%, and META at 81.9%. The actionable point is not that Alphabet has the highest margin, because it does not. It is that Alphabet is delivering +21.8% revenue YoY with a 62.4% gross margin while funding a custom AI infrastructure build, and that combination screens differently from companies where growth is either much higher but more concentrated, as with NVDA at +85.2%, or margin is higher but capex intensity is a more visible investor concern, as with META at 81.9% gross margin and +33.1% revenue YoY.

What would break the thesis is now numerical and near-term. For the next quarter, the first confirmation point is whether revenue can hold near the Q1 FY2026 level of $109,896.0 million while sustaining revenue YoY near +21.8% and gross margin near 62.4%; a retreat toward Q4 FY2025 gross margin of 59.8% without a matching acceleration in Cloud would weaken the operating-leverage case. The second point is whether management keeps 2026 capex within the $175 billion to $185 billion range and links it to backlog conversion rather than only capacity availability; a move above $185 billion without backlog growth beyond $240 billion would shift the debate back to overspend risk. The third point is whether Search growth remains near the disclosed 17% and Cloud growth remains near 48%, because those are the two revenue engines underwriting the AI build. Finally, on the next call, watch guidance_tone versus qa_evasiveness: Q4 FY2025 had guidance_tone of 0.65 and qa_evasiveness of 32.0, while Q1 FY2026 had guidance_tone of 0.28 and qa_evasiveness of -1.5. The best confirmation would be guidance_tone recovering from 0.28 while qa_evasiveness stays far below 32.0, because that would pair cleaner delivery with the financial proof already visible in the print.