How bad does the world have to be before Palantir is happy? In the sunny U.S. tech sector, the software company stood out for its downbeat warnings of global instability. AI may be the crisis it’s looking for.
Now is the high tide of cynics. Artificial intelligence is being touted as a dystopia in the making. In my meetings with tech companies, I keep hearing the phrase “the Oppenheimer moment” — a reference to Robert Oppenheimer, the physicist who led the development of the atomic bomb.
It’s unclear exactly how generative AI will destroy us, or what kind of wealth it will create. But it turns out that all this talk of great power is extremely helpful to the value of a handful of companies. This week, artificial intelligence chip maker Nvidia’s market value briefly topped $1 trillion. AI startups such as Character.ai and Anthropic have continued to raise money even as funding elsewhere has dried up. Palantir’s shares have more than doubled in five months.
This week, Palantir made a broad rollout of its new AI platform. The tool can generate conversational responses using large language models, or LLMs, powering chatbots like ChatGPT. Because it’s based on customer-specific data, it should avoid hallucinations—the wrong answers that plague other chatbots. A demo on YouTube shows how it works on the battlefield, helping to identify an enemy tank and giving advice on how to target it. The Ukrainian military is already using some of its initial capabilities, the company said.
Palantir isn’t the only software company racing to show how generative artificial intelligence can be used more productively than writing college papers. IBM also announced a new artificial intelligence platform called Watsonx. But IBM’s stock price has fallen this year.
Palantir seems to do a better job of illuminating real-world uses than most. “You need a core set of technologies that will allow you to bring these LLMs to your enterprise — to process your data,” says CTO Shyam Sankar. “Then you need a really strong governance control layer that allows you to build AI trust.”
It helps that AI-style puzzles and existential threats are Palantir’s usual approach. It’s hard to imagine a company that talks more about disaster. Last year, it warned that the world was underestimating the threat of a nuclear attack and put it at around 20% to 30%. Co-founder and Silicon Valley investor Peter Thiel is known for making his own chilling statements about global annihilation. In 2008, he described the reemergence of what he called the modern apocalyptic dimension. While the 20th century, he wrote, was “both great and terrible,” the 21st century promises to be both. Fear of artificial intelligence fits perfectly with this worldview.
What does Palantir actually do? The $31 billion company is sometimes described as technology’s answer to management consulting. It was created after the 9/11 attacks to build software that could be used by intelligence agencies to counter terrorism, and then expanded to other government departments and businesses. Its software combs through data, summarizes information, discovers patterns and presents them in a useful and easy-to-understand manner. Its services have helped BP reduce production costs by about 60%, the company said. It is also the frontrunner for new seven-year NHS contracts worth up to £480m.
Over the past few years, however, Palantir has also become a cautionary tale of what can happen when a company known for working in the shadows steps toward the light. Named after the dark, visionary crystal ball in the Lord of the Rings trilogy, it has for a long time practiced the virtue of secrecy. The idea of an impossibly omniscient tech company was irresistible to the media. In 2018, a Bloomberg article claimed the company “knows everything about you.” A few years later, the New York Times asked if it had seen “too much.”
CEO Alex Karp has done a good job of keeping Palantir’s eccentricity in check. He was known to enjoy German philosophy and described his desire to work with creative, “eccentric” people. When I was in the company’s Denver office, I saw a glass case displaying a drab office suit with a “break in case of emergency” sign.
When the company went public in late 2020, some of the glamor was gone. Suddenly, it’s subject to quarterly earnings reports and clichés from investors who want GAAP profits. The awkward truth is that Palantir hasn’t reported an annual profit in 20 years. This year is expected to be the first it breaks that spell.
Interest in artificial intelligence has restored the company to its mystique. Add in newfound profitability, and the result is a stock price rout. This helps the zeitgeist catch up to Palantir’s way of thinking. If AI does leave us forgotten, don’t expect Palantir to be surprised.
elaine.moore@ft.com