Quick Read – Palantir’s 149% U.S. commercial revenue growth and $1.94B total revenue signal an AI stack value shift SpaceX is poised to replicate. – Pleydell-Bouverie argues falling inference costs erode model providers’ moats, rewarding companies that ground AI in proprietary…
ta over third-party apps. – Palantir’s Q2 earnings report gave AI-stack strategists a concrete data point to rally around. Palantir (NASDAQ:PLTR) reported U.S. commercial revenue of $764 million, up 149% year over year, with total revenue of $1.935 billion growing 92.8% and a Rule of 40 score of 155%
For Clare Pleydell-Bouverie, Co-Head of the Global Innovation Team at Liontrust Asset Management, that number is the clearest evidence yet that value in the AI stack is climbing upward from the model labs into the software and data-organization layers, and she argues Elon Musk’s SpaceX is next in line to demonstrate the same dynamic. Speaking on CNBC on August 4, Pleydell-Bouverie framed Palantir as the archetype of the shift. “Enterprises are adopting Palantir as the full stack enterprise AI platform. We’re seeing a shift in value capture across the AI stack going upwards to that software application layer, to those very few companies that are actually converting tokens into revenue,” she said.
CEO Alex Karp echoed the language on the earnings call, calling the quarter “otherworldly” and framing the results as validation of demand for “AI sovereignty.” The Falling-Inference-Cost Thesis Pleydell-Bouverie’s core argument rests on economics. As the cost of inference falls, the competitive moat of pure model providers narrows, and the surplus flows to whoever can package intelligence against proprietary data. “Open source models combined with Palantir’s data architecture and tools are now capable of frontier model performance. If you can ground your AI in your own proprietary data, that is significantly preferable to giving away your data to a third party model app,” she said.