Palantir (PLTR) Gave Back Half of a 9.1% Rally While Snowflake Kept 21%
Palantir closed Thursday at $179.01, up $7.97 or 4.66%. The close understates how the session opened and overstates what it settled into. The stock printed as high as $185.76 and was running 9.1% higher by 10:10 in the morning, then spent the balance of the day returning the gain. It eased again after the bell to $177.80. Volume came in at 22.23 million shares against a ten-day average near 30.8 million, so the bid that produced the best price of the week was thinner than an ordinary day’s trade.
The company put two items on the tape. It expanded its alliance with PwC US to cover enterprise AI, M&A transformation and ERP modernization, pairing Foundry and AIP with PwC’s engineering and managed services organization. It also named a new global head of financial services. Neither disclosure carried a dollar figure. Two days earlier the Army had moved TITAN into production with a $192 million award, $127 million of that to Palantir for eight initial systems over eighteen months, and the stock fell roughly 6% the following session.
Read those two reactions side by side and the conclusion is uncomfortable for anyone pricing this name off contract flow. The award with a disclosed value, a delivery schedule and a named customer moved nothing. The alliance with no value attached moved $19 billion of market capitalisation. That is not a market weighing revenue. It is a market weighing narrative surface area.
The more honest explanation for Thursday sits outside Palantir entirely. Snowflake reported Wednesday after the close and delivered $1.55 billion against a $1.48 billion consensus, adjusted EPS of $0.62 against $0.45, and product revenue of $1.49 billion growing 37%. That was the third consecutive quarter of product revenue acceleration, and the October guide of $1.59 billion implies a fourth. Snowflake opened up 23% and held better than 21% into the close. ServiceNow added 6%. Salesforce added 3%. Oracle added 3%. Palantir added 4.66% and finished at the bottom of the group.
That ordering is the finding. Snowflake’s print is the cleanest available evidence that enterprise AI budgets are still accelerating rather than plateauing, which is the single macro input the Palantir commercial thesis depends on most. US commercial revenue grew 149% last quarter to $764 million and management guided the full year above $3.424 billion. If the cohort re-rates on proof that enterprises are spending, the company with the highest exposure to that spending should lead the move. It lagged by a factor of four and gave back half of what it did capture.
The mechanical reason is multiple headroom. Palantir trades at 144 times trailing earnings and roughly 98 times forward. Snowflake was cheap enough on a growth-adjusted basis for an acceleration surprise to reprice it. Palantir has no comparable slack. Good news that belongs to the category rather than to the company cannot lift a stock that already carries the category’s most expensive assumptions. It only defends the ones already there.
On the competitive question that produced Wednesday’s decline, the moat holds where the analyst note aimed and is quietly weakening somewhere else. Google shipping AI tooling for government does not touch what makes Gotham hard to displace. TITAN is the demonstration. The Army named Palantir prime, with L3Harris, Sierra Nevada, Strategic Technology Consulting and World Wide Technology underneath it, which inverts the usual defence integration hierarchy. What earns that position is an accredited ontology sitting inside a live targeting chain, four years of prototype iteration, and authorities to operate that a model vendor does not hold and cannot quickly acquire. A better model does not clear an ATO.
The PwC alliance is the item that deserves more scrutiny than it received. Palantir’s original defensibility included the forward deployed engineer, the practice that made each deployment specific to the customer and expensive to unwind. Routing implementation through PwC’s managed services converts a proprietary asset into a channel. It raises the growth rate and lowers the barrier at the same time, because PwC implements Snowflake, Databricks and the rest of the field from the same bench. The alliance is bullish for the revenue line and corrosive to the thing that justifies 98 times forward earnings. Both are true, and the market priced only the first one on Thursday, then thought better of it by three in the afternoon.
Positioning. The stock sits 13.7% below the $207.52 record set last November and 68% above the June low of $106.37, up 46% over the past month and roughly flat for the year. Base case is $165 to $195 into the November print, a range the stock has already established and has no obvious reason to leave without company-specific news. Bull case takes out the record toward $220, matching the more aggressive published model targets, and requires US commercial revenue above roughly $900 million in Q3 or a PwC-sourced win with a disclosed contract value. Bear case is cohort derating rather than a Palantir miss: if the enterprise AI software group’s next round of prints fails to sustain the acceleration Snowflake just showed, a compression from 98 times forward toward where ServiceNow and Salesforce trade puts the stock at $120 to $140 without Palantir doing anything wrong. Michael Burry’s renewed short thesis moved to the balance sheet and revenue recognition on Thursday, which is harder to answer with a partnership announcement than a valuation argument is, and it is the kind of line that becomes dangerous only if sell-side picks it up.
The number to watch in November is whether the Q3 disclosure separates US commercial revenue sourced through partners from revenue closed direct. If it does not, there is no way to tell whether the PwC alliance is distribution or the beginning of the deployment layer becoming contestable, and the stock will keep trading on announcements with no figures attached.