Palantir CEO compares excessive AI use to a porn addiction, says more AI use does not always create more value

Palantir CEO Alex Karp said excessive AI use does not automatically create value and compared it to addiction. His remarks reflect a wider industry shift from tracking AI usage to measuring business outcomes.

In Short

• Palantir CEO compares excessive AI use to porn addiction
• Companies question whether rising AI spending boosts productivity
• Palantir says more AI usage doesn't always create value

Palantir CEO Alex Karp has entered Silicon Valley's growing debate over AI usage with a comparison that has raised eyebrows. Speaking during a live interview on TBPN on the sidelines of Palantir's AIP Con 10 event, Karp compared excessive AI use to porn addiction, arguing that simply consuming more AI does not necessarily create more value.

"Really, we call it the demastibatory, like get off masturbation thing internally," Karp said while describing Palantir's approach to token tracking.

"Sure, it's like people are just sitting there all day kind of like a porn addiction," he added.

While the language was unusual, Karp's comments reflect a growing concern among technology companies: rising AI usage is not always translating into higher productivity or better business outcomes.

Why are companies questioning AI usage?

The debate centres around a trend known as "tokenmaxxing", the idea that employees should maximise their use of AI tools and consume as many AI tokens as possible. For much of the AI boom, heavy AI usage was often seen as a positive sign. More prompts, more tokens and more interactions with AI tools were viewed as indicators that employees were embracing the technology.

However, that thinking is now facing increasing scrutiny.

As more AI companies switch to usage-based AI pricing, many have reported rapidly rising AI bills without a noticeable increase in productivity. Uber COO Andrew Macdonald recently highlighted the issue, saying the company was struggling to see a clear link between growing AI spending and meaningful returns. Amazon has also reportedly removed an internal AI leaderboard after employees allegedly began inflating their AI usage.

More tokens means more slop

Karp's comments closely echo remarks made by Palantir CTO Shyam Sankar during the company's earnings call last month. Sankar said Palantir considers itself "a no slop zone" and argued that businesses should not assume that cheaper AI or higher token usage automatically creates value.

"More tokens means more slop," Sankar said. "And the more commodity cognition you consume, the more you need a system that can prevent the economic harm, so you can harness the economic value."

According to Palantir, AI becomes valuable when it is connected to real-world business processes rather than being used simply for the sake of usage.

Where AI works well and where it doesn't

Karp acknowledged that there are many problems AI models can solve effectively. For example, he said AI can easily generate a report on topics such as China's GDP growth.

However, he argued that more complex business challenges require something more than a chatbot response.

Questions involving supply chains, industrial operations, military logistics, manufacturing processes or specialised oil and gas operations require continuous decision-making and precise workflows. According to Karp, these are areas where large language models can enhance human work but cannot replace the underlying processes.

"They are enhanced by large language models. They are not replaced by large language models," he said.

The focus is shifting from AI usage to AI value

Karp said many AI capabilities will eventually become widely available. What will remain difficult to replicate is identifying the right business problems to solve. In his view, that comes down to human judgement and what he described as "taste."

- Ends