Palantir's full-year 2025 adjusted free cash flow was about $2.27 billion. Alex Karp has told investors the company will be producing $15 billion to $18 billion of it within two years.
Most of the argument over the stock is about whether any company can grow fast enough for a decade to deserve the price. Chris Markoch said the math only has to work for two years, and that this is the part of the story the market has not priced.
"Karp saying I got to do it for the next two years and if I do this for the next two years I'm playing with house money for the next eight and that's the valuation that hasn't been repriced in yet."
Markoch writes on stocks for MarketBeat, has been calling Palantir undervalued on the show since January or March, and says he plans to hold the shares through 2027 and 2028.
I listened to the full interview so you can skip it. 24 minutes of audio, 14 minutes of reading.
Here are the 10 takeaways that matter.
👤 Guest: Chris Markoch, a MarketBeat analyst and writer who covers Palantir for the show and says he plans to hold the stock into 2028
🎙️ Host: Bridget Bennett, who presents MarketBeat's stock coverage on YouTube
📰 Published: 9 September 2026 on YouTube (MarketBeat)
🔴 YouTube | 🔗 Episode page | ⏱️ 24 min | ✅ Time saved: 10 min
Key Takeaways
The valuation case rests on two years of cash flow growth, not ten
Karp's target is $15 billion to $18 billion of free cash flow against about $2.27 billion in 2025
The same cash-flow model produces $74 a share at 20% growth and roughly $600 at 50%
The stock was around $173 when the interview was recorded
August's 30% move came from investors treating a good earnings report as good news
Palantir has still lagged: up 9.8% over 12 months, on Markoch's numbers
A $192M Army contract matters for the role it creates, not the revenue it adds
Palantir's share is about $127 million, for eight next-generation TITAN ground stations
The Army is now paying Palantir to coordinate Anduril and L3Harris, not just its own software
Markoch says the criticism that Palantir depends on the US government is the reason commercial customers trust it
A customer that drops Palantir loses what Palantir built, which is why he calls the company irreplaceable
Karp is a salesman and the data-ownership problem is real, Markoch says — both at once
Companies withhold data from general-purpose AI models and hand all of it to Palantir
1. Why the Stock Jumped 30%
Bridget Bennett opened on the move in the shares in early August — "The stock has really moved a ton. It jumped roughly 30% in early August." — and asked what caused it. Markoch put it down to the earnings report and to investors finally reading it the way he thought it should be read.
He said the report removed the fear that growth was decelerating. "Well, Bridget, it came down to the company's earnings reports and investors finally realizing that good news was good news." He described the report as "stellar on every measure."
The move ran from about $120 to about $170. "And so, yeah, the stock went up from 120 when the earnings report came out and now it's up around 170."
He had been calling the stock undervalued and oversold since January or March, and said investors had finally interpreted the results that way and rewarded it.
Markoch tied the reaction to a wider sorting inside the AI trade between the companies that benefit and the ones that do not. His verdict on where Palantir falls: "And Palantir is clearly a winner right now."
2. Still a Laggard on the Year
The conversation immediately conceded the bear case on performance: even after the August jump, anyone who bought Palantir earlier in 2026 has little to show for it.
Measured against the rest of the index, the stock has underperformed. Set beside other companies in the AI trade and in the S&P 500 generally, he said, "Palantir's been a laggard."
His 12-month figure was single digits. "It's, as you said, in the last 12 months, it's only up 9.8%." On the year-to-date number he said, "It's still 3 and a half% for the year."
He accepted the complaint rather than arguing with it. "And so certainly, I can understand that if you were an investor that came late to Palantir, you're maybe feeling a little bit disillusioned."
His own holding period is measured in years. "But I've said repeatedly, I'm looking at Palantir as a stock that I plan on holding not only for the rest of 2026, but through 2027, 2028 and beyond." He added: "And I think the payoff is going to be very good if I continue to do that."
3. Karp's $15B Cash Call
The heart of the episode is a valuation argument, and it turns on one number Palantir's chief executive has put in public.
The target is a step change from where the company is now. "Alex Karp talked about hitting a free cash flow target with Palantir of between $15 billion to $18 billion in the next two years." Against that: "Palantir's full year 2025 adjusted free cash flow was around 2.27 billion."
Markoch's reason for taking the target seriously is that Karp has done this before. He said Karp made a similar call in 2023 about where full-year revenue would land at the end of 2025. "And it turned out that he hit that number almost exactly." His general point: "CEOs, founders have a way of kind of knowing their business better than we do."
If the two years land, the rest of the decade can be ordinary. "But if he hits that, then all of a sudden for Palantir to be worth the valuation it's worth today doesn't really look that unreasonable if Palantir just delivers average performance of a software company in the 8 years after that."
He framed this as the gap between what the market is pricing and what Karp is actually promising. Most investors, he said, assume "Palantir has to have this incredible performance for the next 5 10 years," when the promise on the table covers two.
He insisted the bull case is arithmetic rather than sentiment. "It's not just people who are bullish on the stock because vibes. There's actual numbers behind it that you can sink your teeth into."
The open question the exchange left unanswered: if Karp delivers, is there upside left two years from now, or does the current price already capture all of it?
4. The Model: $74 vs $600
Pressed to put numbers behind the argument, Markoch ran a discounted cash flow model — a standard way of valuing a company from the cash it is expected to generate in future years — at two different growth rates and produced two very different answers.
He explained the tool before using it, telling viewers who might not know it: "A discounted cash flow model for investors that might not be familiar with, it's a kind of a standard tool for valuing a stock based on their future cash that they're expected to generate." He said he was using "a very, very, very basic" version of it.
At 20% growth the model says the stock is expensive. "If you look at Palantir and you say they're going to have a conservative 20% annual cash flow growth rate over the next 10 years, it produces a value of about $74 a share." Of that figure: "That's the number that bears have been sinking their teeth into." He put the market price at the time at "$173 right now."
At 50% growth it says the opposite. "However, if you run that same model at 50% annual growth over the next 10 years, which still is quite significantly below Palantir's current growth rate, the fair value jumps to roughly $600 per share."
The two years of very fast cash flow growth are what pull the ten-year average up to 50%. After them, he said, the company can run at ordinary software-industry rates and still clear the average: "Free cash flow growth in the next two years, then for the next 8 years, even if they're averaging around 20 to 30%, which is more of the software industry's average, they're going to be well above that."
That is the choice he says investors are actually making. "And it's really a question of are you believing in the next two years of the story or are you trying to suggest that Palantir has to have this outsized performance of like 50% every year for the next 10 years." His answer to the second version: "That's not true."
5. What Sovereign AI Means
Bennett asked what the catalyst for that growth is. Markoch's answer was the concept Karp has been repeating since the earnings report, and what it means commercially.
The idea is that the customer keeps its own data, models and computing power. "And the idea of sovereign AI is a government or a company needs to keep control of its own data, its own models, its own compute instead of just handing them over to a third-party vendor. That's Palantir's business case in a nutshell."
Palantir's role is to arrange the customer's data, not to hold it. "Palantir is there to orchestrate, not to own a company's assets."
What the customer buys is the routing of information to the people who need it. "Palantir is just there to orchestrate it to make sure that all the areas of the organization are seeing the relevant data that they need to see so that they get the actionable insights that they are looking to get."
6. The $192M TITAN Contract
The news event behind the episode is an Army award announced in the week before recording.
The award covers eight ground stations and names Palantir as the prime contractor. "The Army Contracting Command awarded Palantir a prime agreement to produce and deliver eight next-generation TITAN ground stations." Palantir is partnering with Anduril on it.
The money is small by Palantir's standards. "The total contract is about 192 million. It's split between Palantir and Anduril, but Palantir is expected to get about 127 million of that." Markoch raised the obvious objection himself: set against a free cash flow target in the billions, $192 million is not a lot of revenue.
He argued the contract's value is positional. "It's entrenching Palantir more as a full battlefield system provider for the US military."
Palantir's software was already embedded before this deal. "We've already talked about in the past that Palantir's AIP or Palantir's foundry, it's already the de facto operating system for the US military."
The new element is that Palantir is being paid to coordinate other suppliers' products. "This is entrenching Palantir more and more into that because this contract is now making Palantir more like a system integrator for not just their own products, but also for coordinating with partners like Anduril and L3Harris." The Army continues to own the underlying data.
7. Salesman or Real Problem
Bennett put the standing objection to Markoch directly: "Is this a real issue or is Alex Karp really just a good salesman?"
His answer was that the two are not in conflict. "I think both are true actually." He said Karp is a good salesman and is also describing a real problem.
He pointed at Microsoft's response as evidence the problem is real. Rivals are claiming to offer the same thing, and in his reading are struggling to: "They're trying to bend over backwards to show that they are actually allowing the client to own their data, but it doesn't really seem possible when you're looking at what they're providing."
The corporate version of the problem shows up as data being withheld from AI systems. "And so I think that's really the issue is in practice, you're seeing a lot of companies not giving the LLM access to all of their data." Markoch's point is that a company willing to expose only part of its data has already conceded there is a risk.
He said Palantir customers behave differently. "Whereas with Palantir, companies are very comfortable with allowing it to have all their data because they're not worried about that data leaving the company."
Bennett said the military version of the argument is the easiest to see. "As the military, you don't want any other country's AI model getting a hold of your information." She added that the same case holds for private companies at a moment when "we are just figuring out AI."
8. The Pentagon as Reference
Bennett asked whether Palantir is ahead of the market in working out where the real risks in AI adoption sit. Markoch's answer inverted the most common criticism of the company.
The dependence on government work is, in his reading, the asset. The standing criticism is that Palantir is "too dependent on the US government." His response: "But I think that's really where Palantir's greatest strength is right now."
The relevant credential is the sensitivity of the data it already handles. "But Palantir cut its teeth with the US government, not just the military, not just the Pentagon, but with different departments within the US government."
That record is what it sells to corporate buyers. "It's basically saying the Pentagon trusts us with this. You can trust us with this."
He rejected the idea that Karp is only promoting what he sells. "But I'm saying I don't think Karp's simply talking his book." The argument Markoch says Karp is making is a track-record argument: "Look at who we've established that track record with. Then ask yourself if you think our products are trustworthy."
On the state of the industry generally, he agreed with Bennett's framing that nobody yet knows what the risks are: "We don't know what we don't know and we're trying to figure that out."
9. What Leaving Palantir Costs
Bennett noted that Palantir's commercial customers tend to stay and expand — "They continue their subscriptions once they're there and they see the product and they use it." — and asked whether "irreplaceable" is the right word. Markoch tested it by asking what happens when a customer leaves.
What Palantir builds does not stay behind. When a company drops the software, he said, "the Palantir ontology goes with it. So it doesn't stay with your company." He added: "If you move away from Palantir, you lose all of that. That leaves your company."
A departing customer therefore has to answer what replaces it. "And I think what many people realize is that there is either A) not an answer to what comes next or B) the answer to what comes next is not going to give you what Palantir gives you."
That is the basis for the word. "They're just simply realizing we need this and we can't do without it and there is no other viable alternative to what we're getting from Palantir."
The Army contract is the same dynamic in a stronger form. Asked how irreplaceability shows up in the new award, Markoch voiced the Army's side of it — it no longer has to run the coordination itself: "We don't have to because we trust you to do that and we trust you're going to do that and still serve our best interests." His verdict: "That's an incredibly powerful position for Palantir to be in."
10. The Moving Goalposts
Asked what all of this means for the two- and ten-year outlook, Markoch closed on the pattern he says the criticism of Palantir has followed.
The complaint has changed direction repeatedly. Two years ago the objection was that the commercial business was not growing; six months ago it was that the government business was at risk; now, with government work growing again, he said the concern has moved back to the commercial side.
His reading of that is not flattering to the critics. "They keep looking for things that aren't true."
He said the reports show both halves of the business growing. "They're growing them at above average rates. And there's really no reason to believe that they're not going to keep growing those."
That is what underpins the long-term call. "And if they do continue growing those, I do think that leads Palantir in that trajectory to being one of those winning companies in this AI growth story a couple years from now."
Bonus Insights
Markoch took responsibility for not making the valuation case concrete the first time he was asked: "Bridget, that's a great question and I take the L on that one cuz I didn't tie that up in a nice bow for investors." The discounted cash flow numbers in section 4 were his second attempt at the same answer
Bennett raised regulation as an open variable in the data-security argument, asking, "Will we have legislation that comes along eventually to help secure more of this information?" Markoch did not forecast any, and the pair left it as a reason the competitive picture is still unsettled
He was careful not to turn the sovereign AI argument into an attack on competitors: "This isn't me trying to bash any of these other companies. The answer is I don't know"
On why the next two years matter beyond Palantir specifically, he said the growth still coming to the AI trade and to the economy over that period is "massive," and that the companies that perform through it are the ones that will define the following decade
Markoch's bottom line is that Palantir's price only requires two exceptional years rather than ten, and that the Army's decision to have it coordinate other defense suppliers is worth more to that case than the contract's own revenue.
Products, Companies & Tools Mentioned
Palantir (The subject of the episode — Markoch holds the stock, and calls its AIP and Foundry software "the de facto operating system" the US military already runs on)
Anduril and L3Harris (The partners Palantir is being paid to coordinate under the TITAN award, which Markoch says turns it into a system integrator rather than a software vendor)
Microsoft (Markoch's example of a rival claiming to offer sovereign AI: "They're trying to bend over backwards to show that they are actually allowing the client to own their data")
If this was worth your time, send it to someone who follows the name.
Get the latest market chatter as it happens: