Google's parent company just did something unusual for a tech giant riding high on AI hype: it told the truth about what this costs and what could go wrong.
Buried in Alphabet's annual financial report filed last week is a section that reads less like corporate boilerplate and more like a confession. The company acknowledged that its AI ambitions could cannibalize its core advertising business, that it might end up with massive excess computing capacity it can't use, and that the "large, long-duration commercial contracts" it's signing to build AI infrastructure come with serious operational and financial risks. To fund this uncertain future, Alphabet is planning a $20 billion bond sale—including a 100-year bond, which is either visionary confidence or a spectacular way to make AI someone else's problem in 2126.
This is the same company that reported $82.28 billion in ad revenue last quarter, up 13.5% year-over-year. They're not struggling. They're hedging.
When asked what worries him most, Google CEO Sundar Pichai didn't mention competition or regulation. He said "compute capacity"—specifically power, land, supply chain constraints, and ramping up infrastructure to meet "extraordinary demand for this moment."
That phrasing matters. "This moment" implies something temporary, a gold rush phase that requires massive upfront investment before anyone knows whether the mine contains actual gold or just very expensive rocks.
Alphabet's capital expenditure guidance for 2026 tops out at $185 billion—more than double what the company spent in 2025. Combined with Microsoft, Meta, and Amazon, the big four are projected to increase capex by more than 60% this year compared to 2025's already historic levels. Hyperscalers are spending approximately $1 trillion on AI infrastructure between 2024-2027, with uncertain return timelines.
They're buying Nvidia chips at premium prices, constructing massive data centers, securing power contracts that strain regional grids, and building networking infrastructure to connect it all. Alphabet's long-term debt quadrupled in 2025 to $46.5 billion. CFO Anat Ashkenazi carefully framed this as "fiscally responsible," which is corporate-speak for "we're borrowing enormous sums and hoping this works out."
Here's the part that should interest every marketer and growth leader: Alphabet explicitly acknowledged in its SEC filing that generative AI could reduce search usage, which directly threatens the company's dominant advertising business. This isn't speculation from analysts or doomsaying from competitors—this is Google admitting that its own AI products might destroy its primary revenue source.
The filing states: "We and our competitors are constantly adjusting to meet this shift and provide new and evolving advertising formats. There is no assurance that we will adapt effectively and competitively to meet this shift, and that such advertising formats, strategies, and offerings will be successful."
Translation: We have no idea if this works, but we're committed now.
Gemini, Google's AI assistant, now has 750 million monthly active users—impressive until you remember that Google Search has over 4 billion users. The company is essentially teaching hundreds of millions of people to bypass the search experience that generates most of its revenue. Gartner's 2025 Digital Marketing Report predicts that traditional search volume could decline by 25% by 2028 as AI assistants handle more queries directly, forcing a fundamental restructuring of digital advertising economics.
Every time someone asks Gemini a question instead of searching Google, the company loses an opportunity to show ads. Google's bet is that it can monetize AI interactions as effectively as search queries—but that business model remains theoretical. Nobody has proven that people will tolerate ads embedded in conversational AI responses the way they tolerate sponsored search results.
The risk disclosure about "excess capacity" deserves particular attention. Alphabet is signing long-term contracts for computing infrastructure based on projected demand that may not materialize. If AI adoption plateaus, if competitors develop more efficient models that require less compute, or if the technology simply doesn't generate the expected revenue, Google will be stuck with extremely expensive infrastructure it doesn't need.
This isn't hypothetical. The crypto mining boom left the world littered with warehouses full of specialized equipment that became worthless when the market corrected. The difference here is scale—crypto mining operations didn't involve $185 billion in annual capital expenditure or 100-year bonds.
Alphabet's SEC filing specifically noted risks of "liabilities and obligations in the event of nonperformance by us, our counterparties, or vendors." That's acknowledging that these massive infrastructure contracts create interdependencies where failure by any party—Google, its data center partners, its chip suppliers—could trigger cascading problems.
Including a 100-year bond in a $20 billion raise to fund AI infrastructure is either breathtaking confidence or spectacular financial engineering. A century-long bond means Alphabet is asking investors to bet on the company's ability to generate returns from AI investments over a timeframe that extends well beyond any reasonable technology prediction horizon.
For context, 100 years ago it was 1926—before computers, before the internet, before semiconductors existed. Alphabet is asking bondholders to trust that AI infrastructure built in 2026 will still be generating value in 2126. That's not financial planning; that's philosophy.
What it really represents is arbitrage: borrowing at today's interest rates (reportedly five times oversubscribed, suggesting favorable terms) to fund short-term infrastructure needs while pushing repayment into a distant future where inflation and growth make the debt trivial. It's financially clever, but it also reveals how uncertain the near-term returns actually are.
If Alphabet—with its cash reserves, technical talent, and market position—is this uncertain about AI economics, what does that tell us about the hundreds of companies making smaller but proportionally similar bets?
For marketing and growth leaders, Alphabet's disclosure should prompt honest questions about your own AI investments. Are you buying AI tools because they deliver measurable ROI, or because everyone else is buying them? Are your vendors being transparent about limitations and costs, or are they selling the same "transform your business" narrative that Google is now quietly hedging against in SEC filings?
The companies winning in AI won't be the ones who spent the most or shipped the fastest. They'll be the ones who understood what they were actually buying, what it genuinely cost (including hidden infrastructure, training, and maintenance expenses), and what realistic value it could deliver.
Alphabet's filing is notable not because it reveals that AI is risky—we knew that. It's notable because it shows that even the companies with the most to gain from AI hype are starting to speak publicly about what could go wrong. When Google files disclosures about cannibalization risk and excess capacity, they're not being pessimistic—they're being realistic about a technology that's simultaneously revolutionary and economically unproven.
The question isn't whether AI will change everything. The question is whether it will change everything profitably, and Alphabet just told us they honestly don't know the answer yet.
Strategic AI adoption requires separating legitimate opportunities from expensive gambles. Winsome Marketing's growth experts help organizations evaluate AI investments with the rigor they deserve, ensuring your technology spending drives measurable results. Learn more about our marketing strategy services and AI implementation consulting.