After a 30% year-to-date drawdown, Monday’s Q1 results are Palantir’s first chance to argue, not assume, that it does not belong in the broader AI software multiple compression.
Until two weeks ago, Palantir was, by some measures, the best-performing large-cap software stock of the AI cycle. It had outpaced Salesforce, Microsoft, Oracle, and Adobe by margins normally reserved for early-stage names.
It had crossed $1 trillion in market capitalisation in late 2025. Its US commercial revenue was, by chief executive Alex Karp’s own framing, on a hyperbolic trajectory.
Then it lost almost a third of its value.
By Friday’s close, Palantir shares were down roughly 30 per cent year-to-date and had shed about 17 per cent in four trading sessions. Bloomberg’s curtain-raiser for Monday’s earnings report framed the moment in unusually direct terms: Palantir is now trying to prove it does not belong in the broader software-stack sell-off that has compressed multiples across the sector since mid-April.
What knocked the stock
The proximate trigger was a now-deleted post by short-seller Michael Burry suggesting that Anthropic was “eating Palantir’s lunch” in enterprise AI. The post landed at a sensitive moment.
Anthropic had recently launched a marketplace through which enterprise customers can purchase third-party Claude-powered software, a move that, in some readings, threatened the middleware layer where Palantir’s Foundry and AIP have built their commercial moat. Investors did not need much convincing.
More substantively, Citi cut its Palantir price target to $210 the same week, describing the move as a recognition that AI software multiples were finally compressing from levels that had been hard to defend on conventional metrics. Palantir’s forward price-to-sales ratio, even after the sell-off, remains north of 50, an order of magnitude above most of its software peers.
The analyst case for Palantir, before the past fortnight, had been that the company’s AIP was different in kind from competing enterprise tooling: deployment-led, embedded inside customer operating models, and inseparable from its government franchise.
The bear case was that the same general-purpose foundation models that powered AIP were now being delivered through cheaper, more direct channels. The market, in April, decided to take the bear case seriously.
Monday afternoon’s Q1 results are now Palantir’s best chance to reset the narrative. The Street consensus is for revenue of roughly $1.54bn, up 74 per cent year-on-year, and adjusted earnings per share of $0.28, more than double the same quarter a year earlier.
US commercial revenue is expected at around $772m, up 94 per cent. Government revenue is expected at around $764m, up 57 per cent. Full-year guidance, currently $7.18bn–$7.20bn, has been a closely watched anchor; any meaningful raise would, in the analyst view, partially neutralise the multiple-compression story.
It is, in other words, a quarter that probably will not be missed on the headline numbers. The question is whether the operating commentary, particularly on net dollar retention in the commercial business, on the durability of recently signed government contracts, and on AIP renewals, gives investors confidence that the differentiation thesis still holds. Options markets are pricing roughly a 10.5 per cent post-earnings move.
The structural argument for Palantir, the one its investor base tends to lean on when commercial momentum wobbles, is its government franchise.
The Pentagon’s Maven Smart System, in which Palantir is the primary integrator, was directed earlier this year to be designated a formal programme of record, locking in long-term funding across all military branches by September 2026.
The company also holds an Army enterprise agreement worth up to $10bn over a decade. Neither of those revenue streams is sensitive in the short term to whether Anthropic launches another marketplace. They are, however, also not the part of the business that justifies a 50-times sales multiple.
That tension, between a low-volatility government core and a high-multiple commercial growth narrative, is what investors have been asked to underwrite. The April sell-off suggests the underwriting standard has tightened.
Palantir’s troubles are not happening in isolation. We has been tracking the broader AI valuation question for some time, and the comparison to dot-com-era CAPE ratios, while overstated in places, is not nothing. Anthropic’s reported $850bn-$900bn valuation talks sit at the top of a market in which mid-tier AI-software names have started to retrace, hard.
The question, for Palantir specifically, is whether its differentiation, deeply embedded customer deployments, government anchor, and a functioning AI platform layer, is enough to insulate it from that retracement.
Karp will, presumably, make the case forcefully on the call. He has done so before. Whether the market hears him on Monday, or hears Burry, will depend on the numbers behind the talking points.
If revenue beats and guidance holds, the narrative resets, possibly violently, in the company’s favour. If anything misses, or if commercial commentary disappoints, the past fortnight begins to look like the start, not the end, of a longer reassessment.
Palantir was, until very recently, the AI cycle’s most confident operator. Monday’s call is the first time in the cycle that confidence has had to be argued, not assumed.
