Google just got a $10 billion option to own a bigger piece of the company that's spending like there's no tomorrow on AI talent.
The Summary
- Anthropic granted Google an option to purchase up to $10 billion in Class N stock, deepening ties between the search giant and the Claude maker
- Personnel costs exploded from $489 million to $1.71 billion in Q3, driven almost entirely by stock-based compensation
- The timing connects two threads: Anthropic is burning equity to compete for talent while giving its largest cloud partner a path to own more of the company
- Dilution concerns are mounting as stock grants become the primary currency in the AI talent war
The Signal
Anthropic's Q3 financials reveal a 250% quarter-over-quarter jump in personnel expenses, with the bulk of that $1.71 billion coming from stock-based compensation rather than cash salaries. This is the real cost of competing with OpenAI, Google DeepMind, and a dozen well-funded AI labs for the same pool of researchers and engineers. When you can't outbid on cash alone, you print equity.
The Google stock option arrangement creates a strategic lock-in that benefits both parties. Google gets deeper ownership in a leading AI company without the regulatory headaches of an outright acquisition. Anthropic gets patient capital and continued access to Google Cloud's TPU infrastructure, which it needs to train frontier models.
"Stock-based compensation is the lingua franca of the AI talent market, but it only works if the equity story holds."
But the math gets uncomfortable fast. If Anthropic is granting $1.2+ billion in stock per quarter to employees while also creating $10 billion option pools for strategic partners, existing shareholders are watching their ownership percentages shrink in real time. The company is either betting on explosive growth that makes the dilution irrelevant, or it's in a prisoner's dilemma where everyone knows the equity flood is unsustainable but nobody can afford to stop first.
Key dynamics at play:
- Employee equity grants create alignment but mortgage future value to current talent retention
- Strategic investor options like Google's convert cloud partnerships into ownership stakes
- Class N stock structures typically carry different voting rights, preserving founder control while diluting economics
The Google arrangement also signals something about Anthropic's competitive position. You don't give your cloud provider a $10 billion option unless you need something beyond market-rate compute contracts. This looks like insurance: Google subsidizes infrastructure costs through favorable pricing, Anthropic compensates with equity upside if Claude becomes a category winner.
The Implication
Watch how other frontier AI labs respond. If Anthropic's model works, stock-based compensation becomes the standard playbook for scaling AI teams without burning through cash reserves. If it doesn't, we'll see a correction where only the labs with the deepest pockets or clearest paths to revenue survive the talent arms race. Microsoft's OpenAI stake suddenly looks cheaper by comparison.
For anyone building in this space, the message is clear: capital efficiency matters more than ever. The companies winning in AI right now are either infinitely patient (Anthropic, with strategic equity deals) or ruthlessly focused on revenue per employee (OpenAI's ChatGPT subscriptions, Perplexity's answer engine). The middle ground just got expensive.