🕶️ $13B to kill your OpenAI invoice

The margin is moving, and the enterprise is liquidating humans to chase it.

Baseten raises billions betting companies will pay for cheaper inference, while Oracle spends billions firing people to fund the stack. It's the same capex story told from opposite ends: one racing toward product velocity, the other converting payroll into GPUs. Both assume AI at scale is an infrastructure game first. What's still open is whether the efficiency arrives before the cash does.

Connect your email, calls, product data and more, and Attio instantly builds your CRM with enriched data and complete context. Whether you’re running product-led growth or enterprise sales, Attio adapts to your unique GTM motion.

Then Ask Attio to plan your next move.

Run deep web research on prospects. Update your pipeline as you work. Find customers and draft outreach emails. Powered by Universal Context, Attio's intelligence layer, Attio searches, updates, and creates across your data to accelerate your workflow.

Ask more from your CRM.

Ask Attio

Two years ago, model access was the bottleneck. Now it's a commodity input, and the margin is migrating to whoever makes inference cheapest.

Baseten just closed a $1.5B Series F at a $13B valuation, led by Sands Capital and Wellington Management. Australia's Blackbird made its largest investment ever. The numbers behind the number: 20x revenue growth in one year. Four raises in 18 months.

The pitch is straightforward. Baseten helps companies deploy and run their own AI models at a fraction of what OpenAI or Anthropic charge through proprietary APIs. As open-source models close the performance gap, the old lock-in playbook—charging a premium for API access to a frontier model—faces structural erosion.

Investors aren't just buying Baseten's growth. They're pricing in a thesis: cost-per-inference is the new moat, not parameter count. The infrastructure layer sitting between raw GPU compute and the application is where the margin lives now.

The founder takeaway is concrete. If you're building on a single model provider without a cost-switching architecture, you're already paying the wrong tax. Investor appetite has decisively moved to the companies making inference cheap, not the ones making models big.

Oracle just cut 21,000 jobs—nearly 13% of its workforce—and said the quiet part out loud: AI adoption caused it. Not "macroeconomic headwinds." Not "strategic realignment." The company's own filing states that AI deployment "has resulted, and may continue to result, in workforce reductions."

The price tag for swapping people for compute is staggering. Oracle posted negative free cash flow of $23.7 billion last fiscal year, cranked capex up 162% to $55.7 billion, and spent $1.8 billion on restructuring—nearly five times the year before. The stock is down over 10% year to date.

This isn't an outlier. Meta shed 8,000 employees in May. Microsoft offered voluntary buyouts to 7% of its U.S. staff in April. The pattern is the same everywhere: incumbents are liquidating human capital to finance AI infrastructure buildouts.

What founders should watch: A wave of experienced cloud and AI talent is hitting the open market under real compensation pressure—that's a window. But any startup whose unit economics depend on stable pricing from these providers burning billions to chase scale they haven't monetized should stress-test those assumptions now.

Scale your growth: Join names like HubSpot, Ahrefs, and Deel. Put your product in front of 85,000+ startup founders. Let's talk.