Palantir’s 93% Revenue Growth Fuels Attack on OpenAI and Anthropic

Palantir reported earnings on August 4, 2026 that sent its shares up 26% the following day. The company posted 93% year-over-year top-line revenue growth, 51% cash flow margins, and a 150% jump in U.S. commercial revenue. CEO Alex Karp used the moment to take direct aim at frontier AI labs like OpenAI and Anthropic, arguing that building model-agnostic enterprise tools is a more defensible business than developing large language models that may ultimately compete with the very customers paying to use them.

What happened

Metric	Result
Share price move (day after earnings)	+26%
Year-over-year top-line growth	93%
Trailing 12-month revenue growth	79%
U.S. commercial revenue growth	150%
Cash flow margins	51%
Price-to-sales multiple (at time of recording)	~60x

Palantir reported these figures after market close on August 4, 2026. According to Motley Fool contributors discussing the results on their Hidden Gems Investing podcast, the earnings beat expectations by a wide margin and the company raised its guidance. The 26% single-day share move stood out even in a busy earnings season where beats have been common.

The 150% U.S. commercial revenue figure was the headline number for analysts. As Motley Fool contributor Travis Hoium noted, the customer count is not growing at that rate, which means existing clients are spending significantly more. That signals real, expanding deployments rather than small pilot contracts that never mature.

Why does Palantir’s commercial growth matter so much?

For years, the knock on Palantir was that its valuation only made sense if commercial revenue could catch up to its government business. The Pentagon, for all its budget size, simply cannot grow fast enough to justify a company trading at 60 times sales. A 150% commercial jump is the market finally seeing the evidence it asked for.

One potential soft spot, according to the Motley Fool discussion: remaining performance obligations (future contracted revenue not yet recognised) were roughly flat. In commercial software businesses, that figure can move around more than in government deals, so it may not be a red flag, but it is worth watching in future quarters.

Karp’s argument against OpenAI and Anthropic

CEO Alex Karp used the earnings call to argue that the frontier model approach, building ever-larger general-purpose large language models (LLMs, the underlying AI trained on vast datasets), carries a structural conflict of interest. His point: labs like OpenAI and Anthropic charge enterprises to access their models, then use the revenue and the customer feedback to build their own competing application-layer tools. Customers are effectively funding products that may one day replace them.

Palantir’s counter-position is what it calls model-agnostic AI. Rather than building or owning a specific LLM, Palantir’s platform is designed to connect to whichever model a customer prefers, wrapping it in workflow, data integration, and security layers that an enterprise actually needs. The business value sits in the orchestration and the data pipeline, not in which model generates the output.

This is a real strategic tension in the market right now. We have covered how open-weight AI models are changing the return-on-investment calculus for enterprise buyers. If models keep commoditising, the layer above them, the orchestration and integration layer, becomes the durable asset.

Our take

Karp’s combative style tends to generate more coverage than the substance of his arguments, but this one deserves a straight read. The conflict-of-interest problem he is describing is real. OpenAI and Anthropic both earn revenue from API access and then build consumer and enterprise products that sit in the same categories their paying customers occupy. Whether that is a fatal flaw or just normal vertical integration is an open question, but it is a legitimate reason for a large enterprise to prefer a neutral orchestration vendor.

For businesses evaluating their own AI stack, the model-agnostic framing is useful regardless of what you think of Palantir’s price tag. Locking your workflows tightly into one proprietary model is a risk. Models change, pricing changes, and the lab’s priorities change. Building on a flexible integration layer, whether through Palantir or through more accessible AI integration services suited to smaller companies, gives you room to swap components without rebuilding from scratch.

The 60x sales multiple is still a hard number to get comfortable with. If revenue growth slows from 93% to even 50%, the math gets uncomfortable fast. But for now, the company is executing at a level that makes the bull case harder to dismiss.

What to do about it

1. Audit your current AI vendor contracts for application-layer tools that compete with your own products or services.
2. Ask whether your AI workflows are tied to a single model or provider, and identify the switching cost if that vendor changes pricing or terms.
3. Evaluate model-agnostic orchestration options at your scale. Palantir targets large enterprises, but the same principle applies to smaller businesses using middleware and integration platforms.
4. Watch Palantir’s remaining performance obligations in the next two quarterly reports as a leading indicator of whether the 150% commercial growth is sustainable or front-loaded.

The practical takeaway: your AI strategy should be built on the data and the workflow, not on the assumption that today’s model provider will always be the right one.

Source: Bing News · OpenAI