Google Spent $35.7 Billion in 90 Days on AI Infrastructure

Alphabet’s Q1 2026 earnings, released on April 29, contain a number that should end the “AI search is a temporary feature” debate that has been running through marketing publications for the past year. Google spent $35.7 billion on capital expenditures in the first three months of 2026 alone. That is more than double the $17.2 billion it spent in Q1 2025, and it represents the largest quarterly infrastructure commitment in the company’s history.

To fund the buildout, Alphabet issued $31.1 billion in senior unsecured notes during the quarter. Companies do not borrow $31 billion to support features they might roll back. They borrow $31 billion to build infrastructure they expect to operate for decades.

Google Cloud revenue grew 63% to $20.0 billion in the same quarter. More importantly, Cloud backlog (the contractually committed future revenue customers have already signed up to pay) nearly doubled quarter over quarter to over $460 billion. That is enterprise customers, not consumers, locking in commitments that depend on Google’s AI infrastructure being there to deliver against.

For anyone running an SEO program, an AI visibility strategy, or a content investment plan, these numbers settle a question that has been distorting strategic decisions for two years. AI search is not an experiment. It is the foundation Google has bet the company on, and the financial commitments are now too large to walk back.

The CapEx number in context

Quarterly capital expenditures of $35.7 billion is a number with very few historical comparisons. For most of Google’s history, the company spent roughly $5 to $10 billion per quarter on property and equipment, including data centers, servers, and network infrastructure. That number climbed steadily through 2023 and 2024 as AI workloads grew. It accelerated sharply in 2025. And it just doubled again year over year heading into 2026.

The Q1 2026 figure puts Google on a pace for roughly $140 to $180 billion in annual CapEx if the rate holds. That exceeds the entire annual revenue of most Fortune 100 companies. It exceeds the GDP of many countries. And it represents physical infrastructure, mostly data centers, power systems, custom chip fabrication, and networking, that takes years to plan, permit, and build.

This is the key constraint Pichai has referenced in earlier interviews. Google cannot simply decide to scale back AI infrastructure even if a future executive wanted to. The buildout is years deep into permitted construction, signed leases, manufacturing contracts for custom TPUs, and energy purchase agreements. The infrastructure being built today reflects decisions made twelve to eighteen months ago, and the infrastructure being commissioned in 2027 and 2028 reflects decisions being made right now.

The $31.1 billion debt issuance during the quarter reinforces the commitment. Alphabet has more than $126 billion in cash and marketable securities. The company did not need to borrow money to fund operations. Issuing $31 billion in long-dated debt to fund infrastructure means locking in capital structure for buildouts that will take years to reach full deployment and decades to depreciate. Apparently the strategy is not “spend until results stabilize and then reassess.” It is “lock in the funding now so the buildout cannot be derailed by future financial conditions.”

What the $460 billion Cloud backlog actually represents

The Cloud backlog figure is the most underappreciated number in the entire earnings release. Backlog refers to revenue that has been contractually committed by customers but not yet recognized. A customer signing a three-year, $50 million Google Cloud contract adds $50 million to backlog the day they sign, and that revenue gets recognized as services get delivered over the contract term.

Cloud backlog growing from roughly $232 billion in Q4 2025 to over $460 billion in Q1 2026 means enterprise customers signed nearly $230 billion in new long-term commitments in a single quarter. That is the most significant enterprise commitment to a single AI infrastructure provider in the history of the cloud computing industry.

These contracts are not speculative. They are signed by enterprise procurement teams who model AI workloads three to five years out, calculate ROI, and commit company budgets to long-term agreements. When $460 billion of those commitments cluster around one provider, the enterprise market has effectively voted on which AI infrastructure they expect to be running their AI workloads on through 2030 and beyond.

For brands thinking about AI visibility, the Cloud backlog matters because it indicates which AI ecosystem will keep absorbing more enterprise content, more enterprise data, and more enterprise queries. Google Cloud customers feed Gemini’s enterprise deployments, train custom models on Google’s infrastructure, and integrate AI agents into their workflows through Google’s APIs. The more workloads concentrate around Google, the more Google’s AI products get exposed to the data, queries, and use cases that shape how those products evolve.

Why this changes the AI visibility conversation

A common reading of AI search through 2024 and 2025 went something like this: AI Overviews might be a temporary experiment Google could pull back if engagement metrics declined. ChatGPT might lose ground if OpenAI’s funding model proved unsustainable. The whole AI search wave might recede if hallucination problems persisted or user trust collapsed. Under those assumptions, investing heavily in AI visibility looked premature. Better to wait for the dust to settle.

The Q1 2026 numbers seem to make those assumptions unsupportable. Google has committed enough capital to AI infrastructure that pulling back is no longer a strategic option. The company is locked into operating this infrastructure at scale for the entire useful life of the assets, which extends well into the 2030s. Whatever AI search looks like over the next decade, it will run on infrastructure Google has already paid to build.

We have to conclude that the planning window for AI visibility work has compressed. Brands assuming they have years to figure out their AI visibility strategy are working from a timeline that no longer matches reality. The infrastructure is being built now, the enterprise commitments are being made now, the user behavior is shifting now, and the brands compounding their citation presence and entity recognition during this window will reach the next stage with momentum that competitors cannot easily replicate.

What the spending tells us about Google’s AI roadmap

Pichai’s earnings statement included a specific data point that helps interpret what the infrastructure is actually being built for. He noted that Gemini is now processing more than 16 billion tokens per minute via direct API use by customers, up 60% from the previous quarter. He also reported 350 million paid subscriptions across Google’s services, with the strongest quarter ever for consumer AI plans driven by the Gemini App, and Gemini Enterprise paid monthly active users growing 40% quarter over quarter.

These numbers describe a stack that is being rebuilt around AI consumption at every level. The consumer side, through the Gemini App and Google’s subscription products, is reaching streaming-service scale. The enterprise side, through Gemini Enterprise and Google Cloud’s AI infrastructure, is consolidating major enterprise workloads. The developer side, through API token throughput at 16 billion per minute, indicates that Google is becoming critical infrastructure for AI-powered third-party applications.

Each of these layers feeds back into search visibility in different ways. Consumer Gemini use shapes which brands users discover through conversational AI. Enterprise Gemini use shapes which brands appear in AI-powered enterprise applications. Developer API use shapes which content gets pulled into the AI-powered tools and products that millions of people use without ever directly opening Google.

The CapEx and Cloud backlog numbers tell us Google is building infrastructure to serve all three layers at unprecedented scale. The implication for content strategy seems clear. Brand visibility now needs to work across consumer AI, enterprise AI, and developer-facing AI simultaneously, because Google is investing to ensure all three remain Google-anchored ecosystems.

What this means for SEO and link building

Three implications follow directly from the Q1 numbers, and they reinforce each other.

The first is that the foundational SEO and link building work has compressed in timeline urgency, not in importance. Authority signals that Google’s algorithm reads, including high-quality backlinks, third-party editorial coverage, structured content, and entity recognition, are also the signals that AI Overviews, Gemini, and Google’s AI infrastructure use to determine which brands to surface. The system is unified at the trust signal layer. Link building and digital PR work continues to compound across both ranking and AI visibility, but the window during which a brand can build that compounding from a competitive starting position is narrowing as established brands lock in their positions.

The second is that the channels Google is investing in (AI Overviews in Search, Gemini App for consumers, Gemini Enterprise for businesses, Cloud-hosted AI for developers) all share the same retrieval logic. Pages that earn citations in one channel tend to earn them in others, because the underlying authority and entity recognition signals are consistent. Guest posting on credible domains and link insertions into authoritative content are not just SEO tactics anymore. They contribute to the citation pool that all of Google’s AI products draw from.

The third is that the “wait and see” strategic posture is no longer defensible on cost-of-capital grounds. Google is committing $35 billion per quarter to make AI search permanent. Companies pricing their content investment as if AI search might not stick are competing against a company that has put $460 billion of contractually committed enterprise revenue against the opposite bet. Apparently, the question has been settled. The remaining question is which brands invest now, while compounding has time to work, and which brands wait until the gaps become uncatchable.

The quarterly CapEx as a confidence signal

Public companies reveal their actual strategic priorities in their balance sheets, not in their CEO interviews. Words are cheap. $35.7 billion of capital expenditure in 90 days is not. When Google spends that much on infrastructure, issues $31 billion in debt to fund the buildout, and signs $230 billion in net new enterprise contracts in a single quarter, the company is broadcasting a confidence level in AI search permanence that no marketing announcement could match.

The Q1 2026 numbers are not the kind of data points that show up once and then fade. The infrastructure being built will be operating for decades. The enterprise contracts being signed will deliver revenue through 2030 and beyond. The debt being issued will be on the books for the full term of the notes. Every quarter from here forward, Google will report numbers that reinforce the same story: the company has bet its future on AI infrastructure, the bet is too large to unwind, and the products built on that infrastructure will define how content gets discovered for the foreseeable future.

For SEO programs, AI visibility strategies, link building budgets, and content investment plans, the Q1 numbers function as both a confirmation and a clock. The confirmation is that AI search is permanent and the work compounds. The clock is that competitive positioning in AI visibility is being established right now, by brands that recognized the direction of travel before the data made it undeniable. The Q1 earnings just made it undeniable.