Why Google's Spending $185B on AI (And Running Out of Cash)

FinanceA.I. Is Getting Expensive, Even For Google

TL;DR

Alphabet is raising $85 billion in equity for the first time in 20+ years to fund massive AI data center buildout, spending $185 billion in 2026 alone, but questions loom about whether AI revenue and customer commitments like Anthropic's can justify these investments.

Key Takeaways

1Alphabet's capex has exploded from $30B annually to $185B in 2026, driven almost entirely by AI infrastructure investments totaling $110B above historical spending levels.
2Google's AI revenue ($80B annualized) comes mostly from selling computing power to external customers like Anthropic rather than its own Gemini models, not yet justifying the $110B already invested.
3Companies like Uber are discovering 'token maxing'—employees consuming unlimited AI tokens without measurable productivity gains—leading to spending cuts and budget caps.
4Anthropic has committed to $70B in annual cloud spending but its revenue sustainability is questionable, raising doubts about whether it can meet its $200B Google commitment.
5Alphabet is raising equity rather than debt to avoid financial risk if the AI bubble bursts and customers like Anthropic default on spending commitments.
6Market sentiment around Google's AI leadership has inflated its stock and enabled cheap capital raising, but leadership likely acknowledges the current spending boom may be irrational.

Notable Quotes

“You make your head explode, right? When you hear companies talking about 25% of code commits over the last quarter were AI driven... but then you talk to your senior engineering leaders and say 'how many projects that were on the cutting room floor got moved above the line because of productivity gains?' That link is not there yet.”
— Andrew McDonald, Uber COO

“Last year at IO that grew to about 480 trillion tokens and fast forward to today that number has jumped seven times to 3.2 quadrillion tokens per month. Never imagined I'd say quadrillion in an IO keynote, but here we are.”
— Sundar Pichai, Alphabet CEO

“Now, some out there might call this token maxing, and there's probably some truth to it.”
— Sundar Pichai, Alphabet CEO

Chapters

1. The $85 Billion Capital Raise

Alphabet announced an unprecedented equity capital raise of $85 billion, its first in over 20 years since its 2004 IPO. The company's internal cash flow of $165 billion is no longer sufficient to cover planned spending, forcing it to issue new stock instead of taking on additional debt.

2. Explosive Growth in Capital Expenditures

Capital expenditures have skyrocketed from $30B annually to $52B in 2024, $92B in 2025, and $185B planned for 2026. Since 2024, Alphabet has spent $110B above historical spending levels, almost entirely on AI data centers.

3. AI Revenue Sources and Profitability

Google generates roughly $80B in annualized AI revenue: $60B from Google Cloud (mostly selling computing to Anthropic) and $20B from subscriptions. With estimated 20% operating margins, Alphabet generates only $16B in annual AI operating profit, requiring 7 years to recoup its $110B AI investment so far.

4. Anthropic's Unsustainable Spending Commitments

Anthropic has signed $70B in annual cloud spending commitments ($40B with Google, $10B with AWS, $15B with SpaceX, $6B with Microsoft). With 2025 revenue near $5B and Q2 2026 projected at $11B annualized, Anthropic must more than double its revenue again to afford these commitments.

5. Token Maxing and Corporate AI Waste

Companies are discovering that surging AI token consumption doesn't translate to measurable productivity gains. Uber burned through its entire 2026 AI budget by mid-May and implemented a $1,500/employee/month cap. Anonymous sources report companies spending $500M monthly on Claude Code, suggesting widespread waste and unsustainable usage patterns.

6. Alphabet's Conservative Hedging Strategy

Instead of taking on additional debt, Alphabet chose to raise equity, limiting financial risk if the AI bubble bursts. If Anthropic defaults on its commitments, Alphabet will have underutilized data centers but won't face bankruptcy from debt obligations.

7. Is the AI Boom Rational?

Alphabet CEO Sundar Pichai previously acknowledged that trillions in data center spending may not be rational and used the term 'quadrillion tokens per month'—hinting at token maxing. Despite this, the company continues aggressive investment because market sentiment rewards AI spending and stock price momentum.

Key People & Entities

Sundar Pichai: CEO of Alphabet; acknowledged the potential irrationality of AI spending while leading the company's aggressive data center buildout
Andrew McDonald: Chief Operating Officer of Uber; discussed the company's AI spending overages and lack of measurable productivity improvements
Anthropic: AI startup that committed to $200B in cloud spending with Google and represents nearly half of Google Cloud's revenue backlog
Alphabet / Google: Parent company planning to spend $185B on capex in 2026, mostly on AI data centers; raised $85B through equity offering

Glossary

Capital Expenditures (Capex): Money spent by a company to acquire or upgrade physical assets such as data centers, buildings, and equipment.
Token Maxing: When employees consume excessive AI tokens (API calls) to appear productive without generating measurable business value, wasting company money.
Operating Profit Margin: The percentage of revenue that remains as profit after covering operating expenses; indicates how efficiently a business operates.
Stock Buyback / Repurchase: When a company buys back its own shares from the market, reducing the total number of shares outstanding and potentially increasing earnings per share.
Equity Capital Raise: When a company raises money by issuing and selling new stock shares to investors, diluting existing shareholders' ownership percentage.
Share Dilution: The decrease in earnings per share and ownership percentage that occurs when a company issues additional shares of stock.
Backlog: In revenue terms, confirmed future orders or commitments from customers that the company will recognize as revenue over time.

Explore