Every gold rush eventually produces the people selling shovels. Alex Karp wants Palantir to be the shovel store.

The Palantir CEO is making a pointed case that companies should stop signing directly with AI model providers like Anthropic and instead route their AI spending through intermediaries, specifically his company. His argument boils down to something most CFOs already suspect: spending enormous sums on AI infrastructure without a clear path to measurable returns is a recipe for regret.

The numbers backing the pitch

Karp’s positioning isn’t just talk. Palantir’s Q1 2026 earnings gave him a strong hand to play. The company reported revenue of $1.63 billion, an 85% jump compared to the same period last year, with adjusted earnings per share landing at $0.33.

US commercial revenue, arguably the most closely watched segment, grew 104% year-over-year. Palantir expects that figure to accelerate to 120% by year’s end.

The company also raised its full-year 2026 revenue guidance, now projecting a growth rate of 71%.

Palantir isn’t selling AI models. It’s selling the layer that sits between enterprises and the models themselves. The company’s Artificial Intelligence Platform, or AIP, is designed to integrate AI capabilities with a company’s existing data infrastructure.

The pricing model reflects this philosophy. Instead of charging based on how much companies use a particular AI model, Palantir ties its fees to outcomes. In English: you pay based on what the AI actually accomplishes for your business, not how many tokens it processes.

Why vendor lock-in is the real concern

Karp’s warning about over-reliance on a single AI provider taps into a growing anxiety among enterprise buyers. When a company goes all-in on one model provider, it’s essentially betting its AI strategy on that vendor’s roadmap, pricing decisions, and continued existence in a brutally competitive market.

Palantir’s pitch is that its platform acts as a buffer. Enterprises can swap underlying models without rebuilding their entire AI infrastructure.

Karp draws a sharp distinction between what he calls results-driven AI deployments and broader, less targeted applications. The first category delivers tangible business outcomes: faster decisions, reduced costs, better predictions. The second category is where companies tend to hemorrhage money, deploying AI because it sounds innovative without a clear connection to the bottom line.

What this means for investors

Palantir’s 85% revenue growth and raised guidance signal that enterprise demand for its intermediary approach is not just holding steady but accelerating. The US commercial segment’s trajectory from 104% to a projected 120% growth rate is particularly notable, suggesting American companies are adopting the platform at an increasing clip.