Yet Palantir keeps surprising people. You’ve probably heard the hype, seen the short squeezes, watched the stock climb from $10 to over $90 in five years. It feels like a story stock—one built on vision, controversy, and a cult following. Nvidia, meanwhile, prints money. Its data centers power nearly every major AI model on the planet. So why are some investors still asking if Palantir could outperform?
The Nvidia Juggernaut: More Than Just a Chip Company
Market cap over $2.2 trillion. That number still stuns me. For context, that’s more than the entire S&P 500 had in 1980. And it’s not just size—Nvidia’s growth is accelerating. Revenue jumped 126% year-over-year in Q1 2024, fueled almost entirely by data center demand. This isn’t speculative. It’s not based on future promises. It’s happening right now, in real time, on balance sheets audited by Deloitte.
How Nvidia Became the Engine of the AI Revolution
It started with GPUs—graphics processing units originally built for gaming. But their parallel computing power turned out to be perfect for training neural networks. While Intel and AMD stuck with CPUs, Nvidia pivoted hard into AI. They didn’t just sell chips—they built an entire ecosystem: CUDA software, DGX servers, AI enterprise tools. That moat is miles wide. Recreating it would take billions and a decade. The thing is, no one’s even close.
And that’s why Microsoft, Google, Meta, and Oracle are lining up to buy every H100 they can get their hands on. Each H100 sells for about $30,000. A single AI training cluster might use 10,000 of them. So we’re talking $300 million per installation—just for the chips. Multiply that across dozens of data centers worldwide, and suddenly, Nvidia’s $30 billion quarterly revenue starts making sense.
Nvidia’s Financial Muscle: What the Numbers Reveal
Last year, gross margins hit 76%. That’s Apple-level profitability, but in hardware. Even more telling? Their backlog exceeds $27 billion. Customers aren’t just interested—they’re begging to pay. Taiwan Semiconductor Manufacturing Company (TSMC) can’t fabricate chips fast enough. Nvidia can’t ship them quickly enough. This is a demand-constrained business, not a supply-constrained one. That changes everything.
But here’s where it gets tricky. At a forward P/E of about 30, Nvidia is not cheap. If AI adoption stalls—if hyperscalers pause spending—the stock could correct hard. Yet, given that global AI investment is projected to reach $1.3 trillion by 2027 (IDC data), that risk seems priced in. Or at least, discounted. You’re paying for perfection. But so far, they’re delivering it.
Palantir: The Stealth Data Powerhouse You’re Underestimating
Let’s be clear about this—Palantir is not a tech darling because of its product demos. It’s because of its clients. The CIA. The U.S. Army. The NHS in the UK. BP. Airbus. These are not early adopters. They’re risk-averse institutions that don’t bet on vaporware. And yet, they’ve signed multi-year contracts worth hundreds of millions. That’s validation on a scale most startups dream of.
Palantir’s Two Sides: Gotham and Foundry Explained
Gotham is their defense and intelligence arm. It digs through petabytes of classified data to find patterns—terrorist networks, fraud rings, supply chain breakdowns. Foundry is the commercial version. It helps companies like Merck or Coca-Cola optimize manufacturing, logistics, and R&D. They’re different beasts, but both rely on the same core: ontological modeling. That’s a fancy way of saying they map relationships between data points so machines can “understand” context—not just process information.
It’s a bit like teaching a computer to read between the lines. And that’s exactly where Palantir claims its edge. While others throw more data at models, Palantir insists on structuring it first. Skeptics say that’s outdated. But when lives or billions in revenue are on the line, accuracy matters more than speed. This philosophy has led to a gross retention rate of 115%—meaning existing customers spend more over time. That’s rare in enterprise software.
Palantir’s Profitability Turnaround: From Burn Rate to Bottom Line
Five years ago, Palantir lost money on every dollar of revenue. Today? Net margins are over 20%. That turnaround wasn’t luck. It came from brutal cost discipline, automation, and a shift to cloud-based deployment. Their commercial segment grew 34% year-over-year in 2023, while government revenue climbed 22%. Not Nvidia-level, sure. But sustainable. And crucially, they’ve achieved this with just 1.5% of AWS’s market share.
Because here’s the irony: Palantir runs largely on AWS. They’re a customer of the very cloud giants they compete with. That creates tension—but also flexibility. They don’t need to build data centers. They focus on software, integration, and domain expertise. It’s a capital-light model, which makes scaling easier. And in uncertain economic times, that’s not nothing.
Nvidia vs Palantir: A Tale of Two Growth Profiles
Let’s compare them side by side—not just financially, but strategically. Nvidia is a horizontal play. Their chips go into everything: cloud servers, self-driving cars, robotics, AI PCs. Palantir is vertical. They dominate specific use cases in defense and industrial analytics. One is a shovel seller in a gold rush. The other is a specialist miner with proprietary tools.
Revenue and Margin Comparison: Scale vs Precision
Nvidia pulled in $60.9 billion in revenue for FY2024. Palantir? $2.3 billion. That’s a staggering gap. Even their growth rates—126% vs 20%—tell very different stories. But margins are another story. Palantir’s adjusted operating margin is 31%. Nvidia’s is 56%. Both are excellent, but Nvidia’s scale magnifies every percentage point. A 1-point margin improvement for Nvidia is worth more than Palantir’s entire annual profit.
Valuation: Premium Pricing and What It Implies
Nvidia trades at around 30x forward earnings. Palantir at about 65x. On the surface, that makes Palantir look overvalued. Except that’s normal for high-growth SaaS companies. Salesforce once traded at 100x. The issue remains—can Palantir grow fast enough to justify that multiple? At current rates, it would need to 3x revenue in four years without multiple contraction. Possible? Yes. Likely? We’re far from it.
Frequently Asked Questions
Can Palantir Compete With Big Tech in AI?
Not head-on. They’re not building LLMs like Google or Meta. Instead, they’re applying AI to structured enterprise data—something big tech often ignores. Their AIP platform lets analysts query databases in plain English. It’s narrow, but powerful in context. Honestly, it is unclear whether this niche can expand beyond regulated industries. But in sectors like defense or pharma, that’s where the money is.
Is Nvidia Overdependent on China?
They used to be. But after U.S. export restrictions on advanced chips, Nvidia adapted. They created China-specific chips like the H20, which comply with regulations but still generate revenue. China now accounts for about 15% of data center sales—down from 25% in 2022. That’s risk reduction, not elimination. Yet, geopolitical tensions could still disrupt supply chains. The long game? Diversify. And they’re doing it.
Will Palantir Ever Become a 0 Billion Company?
They already are. Market cap crossed $90 billion in early 2024 and has flirted with $100 billion since. The harder question is whether they can become a $200 billion company. That would require doubling commercial adoption or landing a few monster government contracts. Possible? Sure. But it would also require flawless execution—and a lot of luck.
The Bottom Line: Which One Should You Buy?
I am convinced that Nvidia is the superior long-term holding. Not because Palantir is weak—but because Nvidia is operating at a different level of technological necessity. Every major AI advancement today runs on Nvidia silicon. That gives them pricing power, ecosystem control, and unmatched visibility into future demand.
But—and this is a big but—if you already own Nvidia, adding Palantir isn’t crazy. It’s a hedge. A bet on domain-specific AI, on government spending, on a team that’s beaten expectations for years. It’s riskier, sure. But because it’s smaller and less efficient, it has more room to improve. That’s the appeal.
To give a sense of scale: if you’d invested $10,000 in Nvidia five years ago, you’d have over $100,000 today. With Palantir? You’d have about $45,000. Not bad. But not life-changing. And that’s exactly where the decision crystallizes. Do you want explosive, infrastructure-level growth? Or a speculative play on data refinement in high-stakes environments?
I find this overrated: the idea that Palantir will “disrupt” Nvidia. They’re in different leagues. What matters more is your time horizon. If you’re investing for 10 years, Nvidia is the obvious choice. If you’re trading, Palantir’s volatility offers opportunities. (Though good luck timing those spikes.)
Experts disagree on how long Nvidia’s lead will last. Some say three years. Others say a decade. The problem is, even when competitors catch up—AMD, China’s Huawei—Nvidia will have moved further ahead. They reinvest 25% of revenue into R&D. That’s $15 billion this year alone. That changes everything.
So here’s my personal recommendation: overweight Nvidia, underweight Palantir. Not because Palantir lacks merit—but because Nvidia is simply harder to bet against. Its technology is foundational. Its financials are bulletproof. And the AI wave isn’t slowing down. In short, if you only buy one, make it Nvidia. The others are just echoes.