A $270 billion software company that almost nobody on Main Street understands. That's Palantir in 2026, trading at 52 times sales with a valuation that would make most portfolio managers spit out their bourbon.
But here's the thing nobody's pricing correctly. Alex Karp's crew didn't just pivot away from bleeding money. They ran the most dramatic margin expansion in enterprise software history, cranking operating margins to 46 percent in three years flat.
Now they've got an AI platform called AIP that's eating the commercial world alive. Government contracts laid the foundation. AIP is the accelerant. The question is whether Main Street can stomach what's coming next.
Let's ground this in reality. Palantir posted $4.48 billion in 2025 revenue, up roughly 56 percent from the prior year. Net income hit $1.63 billion three years after losing $374 million. That's not a turnaround. That's a resurrection.
Free cash flow ran at $1.75 billion. EBITDA cleared $2 billion. Gross margins sit at 84 percent. They're only 4,395 employees doing what legacy defense contractors need 40,000 people to attempt.
The numbers tell the story on their own. Revenue grew 85 percent year over year. Operating margins expanded to 46 percent while rivals were begging for any growth. TTM revenue sits at $5.22 billion with Q1 2026 alone printing $1.63 billion. Forward PE sits at 54x. The stock's at $112.93 but had traded above $200 just months back, so the market is screaming two completely different stories at once.
The magic isn't the AI itself. Every software company is stapling a chatbot onto their product now. It's the ontology. Palantir maps how your data actually relates - factories, suppliers, inventory, customers, shipments, regulatory filings - into a semantic layer that machines can reason over without losing the human logic that built the data in the first place.
Drop a large language model on top of that ontology and suddenly the AI isn't hallucinating. It's reasoning over your actual business. Competitors like Snowflake, Databricks, and C3.ai sell you data pipelines and let you figure out the rest. Palantir sells you the brain. That's the difference between a tool and a moat, and it's why customers don't leave.
Compare this to Lockheed or Raytheon, locked into cost-plus contracts where Congress decides their margins. Palantir doesn't have that problem. They built a platform business model on top of a government foundation, and now they're reaping the pricing power that only software monopolies ever get to enjoy.
The government side is a masterclass in durability. U.S. Army TITAN, the UK's NHS, classified intelligence programs, every Five Eyes nation you can name. These aren't one-time deployments. They're decade-long relationships that compound year after year. Roughly 55 percent of revenue still flows from government, and those contracts pay 80 percent gross margins the entire ride. You can see the trajectory laid out in their investor relations filings.
But commercial is where the fireworks are. Enterprises are finally figuring out that AIP lets them deploy LLMs on proprietary data without giving their secrets to a third-party model first. For most Fortune 500s, that single architectural promise is the whole ballgame.
The adoption curve is still early. Most large enterprises have only run pilots. Once they see what a production LLM can do over their own operational data, expansion revenue compounds like clockwork. That's why revenue went from $884 million in Q1 2025 to $1.63 billion in Q1 2026 - nearly doubled in a single year. Sequential growth of 16 percent quarter over quarter in a $5 billion revenue base is not normal. This is exceptional.
Bears aren't wrong to flinch. The stock's cut nearly in half from its $207 peak. Trailing PE of 127x reads as delusional by any legacy software framework. If AI hype cools even slightly, multiple compression could be brutal. Growth deceleration from 85 percent to, say, 40 percent would send these shares back toward $60. That math is ugly and it's honest.
Bulls see something else entirely. Forward PE of 54x on a business growing 85 percent gives you a PEG barely above 0.6, which is cheap for this kind of margin expansion. Margins are climbing, not contracting. AIP is still in early innings for global enterprise adoption. The total addressable market is literally every company with a data problem, which is every company that exists. That's not a niche. That's the entire economy.
Karp's always been a bit of a madman. But this time his madman thesis is printing cash at 44 percent net margins and the balance sheet looks like a completely different era of the company. Whether that justifies $270 billion is the call you have to make yourself. Just don't call it simple.
Disclosure: The Signal holds no position in PLTR. This is not financial advice.
