_edited.png){kind=logo}
【V+ Perspective】Why Does Palantir Win? In the AI Era, the Scarcest Resource Is Not Technology, But Understanding.
Over the past few months in our monthly meetings with startups, we've noticed something interesting: demos are getting faster. A feature that used to take a quarter now shows up finished by the next monthly meeting — AI has made "building things" easier than ever. But another number hasn't kept up: orders aren't closing any faster.
More things get built. Not more things get sold. That gap is worth every founder's pause.
What's Scarce Has Changed
Why? Our take: AI has moved the location of scarcity.
Engineering used to be the moat — shipping a stable product already put you ahead of most. Now that AI has collapsed the cost of development, whatever you build today, your competitor can build next week. Product advantage is depreciating faster than ever.
What has become scarce is the understanding of customers — the pain points nobody articulates, the real anxiety behind a report, the one department that will stall the rollout. None of this lives in documents. It lives on-site, and even AI can't retrieve it, because often the customer can't articulate it themselves.
The barrier to building is falling. The barrier to understanding is rising.
The Giants Have Voted: The Rise of the FDE
Silicon Valley's giants have already voted with their actions. Look at the hottest job in AI right now: it isn't research scientist — it's Forward Deployed Engineer (FDE), engineers who don't sit at HQ writing code, but embed in the customer's office and grow the product into the customer's workflow.
OpenAI and Anthropic are hiring FDEs aggressively — deploying their most expensive engineering talent to do one thing: understand customers. Google went further, launching vertical Gemini Enterprise offerings for legal and financial services this August — packaging "industry understanding" itself into an enterprise product.
Three months ago we wrote that AI's value lies not in the model but in deployment. Now even the starting point of building has moved into the customer's site.
Palantir: A Decade of Ridiculed "Dumb Work"
And Palantir — the inventor of the FDE model — is the proof.
Over a decade ago, Palantir stationed engineers inside customer sites for years at a time. Silicon Valley mocked it as "a consulting firm in tech clothing" — ugly margins, impossible to scale. In 2023, before its enterprise AI platform launched, revenue grew just 17% and the skeptics were loud.
Then all that customer understanding, earned one deployment at a time, got monetized at once in the AI era. In Q2 2026, Palantir posted $1.94 billion in quarterly revenue, up 93% year-over-year; US commercial revenue grew 149%, with a record $2.13 billion in US commercial contracts booked in a single quarter; full-year guidance was raised to $8.15 billion, an 82% growth rate — the fastest-growing large software company in the world.
Everyone has models. Few understand the enterprise floor. The "dumb work" once ridiculed is now the playbook everyone copies.
Products depreciate. Customer understanding compounds. Palantir spent a decade proving the second half; the AI era will accelerate the first.
Secondhand Understanding Cannot Build a Firsthand Product
So what should founders do? We believe it comes down to one thing: real needs cannot be obtained secondhand. What sales reports back, what consultants summarize, what decks describe — all secondhand. Needs must be experienced with your own senses to be firsthand.
Face these three questions honestly:
1. When did you last sit beside a customer and watch them use your product — yourself, not through a sales rep's retelling?
2. Have you ever run the customer's old workflow with your own hands? Only after being tortured once by that spreadsheet, that approval chain, do you truly know what problem you're solving.
3. Are the top three items on your roadmap conceived in a conference room, or grown from the field?
Conclusion
Our own habits in evaluating deals are changing too. We used to spend a lot of time on the product itself. Now product completeness tells us little — AI can help any team catch up. What we want to know is: how close is this team to its customers? How often does the founder go on-site?
This is also why we often say: thirty percent of our time goes to selecting deals, seventy percent to building them — and building a deal usually means going back to the customer's site together with the team. There is no shortcut to understanding customers. You have to earn it in the field.
Immersion is where building begins. When building gets easy, understanding customers is the deepest moat.
Follow us for more insights:
• LinkedIn: https://www.linkedin.com/company/ventureplusco
About VENTURE+
VENTURE+ specializes in SaaS and AI investments, offering more than just funding. We provide startups with strategic guidance, corporate partnerships, and capital market planning. We aim to be the "Best Co-Founding Partner" bridging startups, venture capital, and industry leaders in long-term collaboration.
Comments