Descriptions:
Palantir CTO Shyam Sankar joins Bloomberg Technology at the Hill and Valley Forum to discuss the company’s Pentagon relationships, its stance on the Anthropic-DOD controversy, and how it thinks about deploying AI across government and enterprise clients. Sankar describes Palantir’s approach to foundation model selection as treating LLMs as ‘commodity cognition’ — maintaining active integrations with all major frontier labs including Anthropic, OpenAI, and others, with the ability to migrate complex government workflows within weeks as the model landscape shifts.
On the Anthropic situation, Sankar was unambiguous: AI founders should not be the arbiters of how military technology is used. He praised Dario Amodei for having principles but argued they were the wrong principles — that decisions about battlefield AI belong to elected representatives and the Department of Defense, which already has policy frameworks like DOD 3000.09 governing autonomous systems. The controversy, he suggested, created an opening that Palantir and others moved to fill.
Sankar also expanded on Palantir’s ‘forward-deployed engineers’ model, describing it as the key to unlocking real economic value from AI — embedding engineers directly with customers to build AI tools around specific workers rather than selling abstract capabilities. He cited a submarine-parts manufacturer that cut two weeks of production planning to ten minutes using AI, and ICU nurses at Tampa General gaining more bedside time. His broader thesis: AI is the antidote to the twentieth century’s managerial overhead, and its biggest beneficiaries are vocational workers with deep domain knowledge.
📺 Source: Bloomberg Technology · Published March 24, 2026
🏷️ Format: Interview
