Palantir Gives Back the PwC Rally as a $27M Sale Moves a $418B Company

Palantir traded at $174.33 on Tuesday, down $8.20 or 4.49% from Friday's $182.53 close, with a session range of $173.67 to $182.19. Volume ran 27.96 million shares against a 30.99 million daily average. Market capitalisation stands at $418.88 billion.

The stock opened near the top of its range and sold off steadily through the session, finishing 4.3% below the high and just 0.4% above the low. After-hours trade added a further $0.09 decline to $174.24.

Nothing operational happened. No guidance change, no contract loss, no downgrade cycle. The identifiable catalyst was a regulatory fund disclosure showing that Cathie Wood's ARK Invest cut its Palantir position, selling roughly $27.2 million of stock while buying $27.5 million of SpaceX.

A $418.88 billion company losing $18.8 billion of market value on a $27 million fund reallocation tells you everything about how this register is constructed. The marginal holder is momentum-driven and reflexive, and the news that moves the price is increasingly about who owns it rather than about what it earns.

That is a change in character from a week ago. On September 3, Palantir jumped 7.7% to $182.53 on an expanded alliance with PwC US — its best session in a month — reversing a 5.8% decline the previous day. Two sessions, a 13.5% round trip, driven by a partnership announcement and a short-seller's social media post.

The broader tape did not help. The Dow Jones Industrial Average (DJI) shed 570.78 points to 52,843.47 while high-multiple software took the deepest damage across the market — Shopify off 7.10%, Guidewire off 7.38%, UiPath off 7.00% — as the 10-year Treasury held 4.80% and the 30-year 5.27%.

Palantir sits at the extreme end of that cohort. At 51 times forward sales, it carries more duration risk than any large-cap software name on the board, and a rising discount rate hits it hardest.

The 52-week range spans $106.37 to $207.52. Tuesday's print sits 16% below the high and 64% above the low.

From $123.06 to $186.38 to $174.33 in Six Weeks

The recent trajectory is violent in both directions and explains why positioning has become the dominant variable.

Palantir closed July at $123.06. It finished August at $186.38 after gaining 51.45% during the month — the best-performing stock in the entire technology sector. It now trades at $174.33.

That is a 51% rally followed by a 6.5% give-back inside six weeks, in a stock with a $418.88 billion market capitalisation.

The first half of 2026 ran the other way. Palantir spent that period as a cautionary tale for growth investors, with a stretched valuation and the broader market rotating away from the theme. By early August the stock sat roughly 40% below its high, consolidating inside a symmetrical triangle near $123.

Year to date the picture is far less dramatic than the August move suggests. As of early September, Palantir was up roughly 2.7% for 2026. A 51% August was largely a recovery of ground lost between January and July rather than fresh appreciation.

That detail matters for anyone anchoring on the August performance. The stock has not compounded through 2026 — it collapsed and then bounced, and the current price sits close to where it started the year.

The August catalyst was earnings. Palantir reported second-quarter results on August 3 after the close, and the stock surged 29% the following session to $162.66, its strongest day in six months. It then continued higher through the month on follow-through buying rather than on new information.

The technical structure now shows the stock below its August close with the $182.53 level acting as the immediate reference. A 4.49% single-session decline on below-average volume — 27.96 million against a 30.99 million average — is drift rather than distribution, which argues the move has more to do with the absence of buyers than the presence of aggressive sellers.

Options positioning has been notably heavy in this name throughout the period, and implied moves have run above 10% around events.

The Q2 Print: 93% Growth and Four Records at Once

The operating results are where both the bull and bear cases start, because nobody disputes them.

Second-quarter revenue reached $1.935 billion, up 93% year over year — the strongest quarter in company history. Adjusted earnings came in at $0.41 per share against a $0.34 consensus, a 19% beat.

Four separate metrics printed all-time highs in the same quarter. Operating margin reached 62%. Net dollar retention hit 157%. Free cash flow margin came in at 63%. US commercial revenue growth ran 149%.

US government revenue grew 90% over the same period.

Guidance moved with the results. Full-year revenue guidance was raised to $8.15 billion to $8.16 billion, with third-quarter revenue guided to approximately $2.16 billion. That third-quarter figure implies an annualised run rate of $8.64 billion, above the full-year guide, which means the exit rate into 2027 is materially higher than the reported 2026 total.

Trailing EBITDA stands at $2.66 billion with an EBITDA margin of 32.18%.

Company filings and releases are published at investor.palantir.com.

A 93% growth rate at a $1.9 billion quarterly revenue base is genuinely rare. Companies of that scale typically decelerate; Palantir accelerated. The combination of that growth with a 62% operating margin and a 63% free cash flow margin is the specific thing that has justified the multiple to holders.

The bear response is not that the numbers are wrong. It is that the numbers, however good, do not support the price — a distinction that makes this debate harder to resolve than a typical growth-stock argument, because there is no data point coming that settles it.

Next earnings arrive November 9, 2026. Between now and then, the stock trades on flow and narrative.

US Commercial at 149% Is the Number That Actually Matters

Strip the segments apart and one line carries the entire investment thesis.

US commercial revenue grew 149% year over year in the second quarter. That is the segment the bull case has always rested on, because government revenue — however large — carries a ceiling defined by appropriations and procurement cycles, while commercial software has no structural cap.

Palantir spent a decade as a government contractor with a commercial ambition that repeatedly failed to materialise. The Artificial Intelligence Platform changed that, and 149% growth is the evidence.

The mechanism is the deployment model. Palantir runs bootcamps that put its platform into a customer's environment for a short engagement, then converts the ones that produce measurable results into contracts. Sovereignty-focused bootcamps recently drew nearly 200 organisations, which is the top of a funnel that has been converting at unusually high rates.

Government revenue at 90% growth is not weak by any standard. A new US Army contract landed in early September, adding to a book that includes the intelligence and defence work the company was built on.

The segment mix is what determines the multiple. A company deriving most of its growth from government work trades at defence contractor multiples — high single digits to low teens on earnings. A company growing commercial software at 149% trades at software multiples. Palantir is currently valued as the second while still deriving a substantial share of revenue from the first.

Which side that resolves toward is the single largest determinant of where the stock trades in two years.

The four platforms — Gotham, Foundry, Apollo and the AI Platform — serve different parts of that split. Gotham is the government product. Foundry is the commercial data platform. Apollo handles deployment across cloud environments. AIP is the layer that has driven the commercial acceleration.

Employee count sits at 4,430, which for a company guiding to $8.15 billion of revenue works out to roughly $1.84 million of revenue per employee.

Net Retention at 157% and What It Actually Measures

The 157% net dollar retention figure is the metric that separates Palantir from every other enterprise software company at scale, and it deserves precision.

Net dollar retention measures what the existing customer base spends this year versus last year, including expansion, contraction and churn. A reading of 157% means the same set of customers spent 57% more, net of everyone who reduced or left.

For context, best-in-class enterprise software companies run 110% to 130%. Anything above 140% is exceptional. At 157% and rising to an all-time high, Palantir's existing customers are expanding faster than most companies acquire new ones.

That figure is the strongest single argument against the bear case, because it demonstrates the product is being consumed more heavily over time rather than being deployed once and abandoned.

It is also where the ambiguity sits. Net retention counts contract value, and contract value in a deployment-heavy model can include implementation and services work that does not carry software economics. A company whose expansion revenue comes disproportionately from headcount-based delivery would show high net retention and low incremental margin.

The reported 62% operating margin argues against that reading. Delivery-heavy revenue does not produce 62% operating margins at scale, and the margin printed an all-time high in the same quarter the retention figure did.

The free cash flow margin at 63% supports it further. Cash conversion above operating margin indicates favourable working capital dynamics, not a business burning cash on delivery.

The counterargument focuses on the balance sheet rather than the income statement, and that is where the short thesis lives.

For modelling purposes, 157% net retention on a $8.15 billion base implies that the existing customer book alone contributes several billion of incremental revenue before any new logo is signed. That is the compounding engine the multiple is paying for.

The question is how long a number like that can persist. Net retention above 150% is mathematically difficult to sustain because the base it is calculated against grows every year.

Sixty-Two Percent Operating Margin on 4,430 Employees

The margin structure is the most underappreciated part of this business and the hardest thing for competitors to replicate.

A 62% operating margin at an all-time high, on revenue growing 93%, means Palantir is capturing operating leverage while investing. Most software companies at that growth rate run negative or low-single-digit operating margins because they spend the incremental revenue on sales capacity.

The headcount explains part of it. At 4,430 employees generating a run rate above $8 billion, Palantir operates at revenue per employee levels that most enterprise software companies do not approach. That reflects a product-led deployment model rather than a large field sales organisation.

Free cash flow margin at 63% — above the operating margin — indicates the business collects cash faster than it recognises revenue on an accrual basis, or that non-cash charges are substantial.

Stock-based compensation is the non-cash item at the centre of the dispute. Heavy equity compensation flatters cash flow while diluting shareholders, and it is one of the three specific accounting concerns raised in the public short thesis.

That is a legitimate issue rather than a technicality. A company reporting 63% free cash flow margins while issuing significant equity to employees is transferring cost from the cash flow statement to the share count, and any valuation built on unadjusted free cash flow overstates what accrues to existing holders.

EBITDA at $2.66 billion against a $418.88 billion market capitalisation puts the enterprise value to EBITDA multiple above 150 times. That is the number that makes the margin story insufficient on its own.

Margins this strong at growth this fast are the reason the stock deserves a premium multiple. They are not the reason it deserves this specific multiple, and conflating the two is the most common error in bullish coverage of this name.

The PwC Alliance and the Consulting Question It Raises

The September 3 announcement was the most consequential news of the past week, and it cuts both ways.

Palantir announced an expanded alliance with PwC US centred on an AI-native deals platform, with PwC deploying Palantir's AI analytics to transform client operations in mergers and acquisitions and enterprise resource planning modernisation. The stock rose 7.7% to $182.53 on the day.

The bull reading is straightforward. A top-tier professional services firm embedding Palantir into its client delivery creates a distribution channel that reaches thousands of enterprises Palantir's own sales organisation would take years to touch. Channel partnerships of that kind are how enterprise software scales beyond the founder-led sales motion.

The bear reading is the timing and the substance. The announcement landed within hours of a widely circulated short thesis characterising Palantir as "a consultant riding a bubble of AI FOMO demand." An alliance with a consulting firm, announced as a rebuttal to a consulting critique, does not fully answer the critique.

The distinction that matters is who does the work. If PwC deploys Palantir's software and Palantir collects licence revenue, that is software economics and the partnership is unambiguously positive. If the engagements require substantial Palantir delivery resources alongside PwC's, the revenue carries services characteristics regardless of how it is labelled.

The 62% operating margin is the empirical answer, and it argues for the software reading. Services businesses do not produce those margins.

Two corporate wins landed in the same window — the Army contract two days earlier and the PwC alliance — which is why the bear case landed without the stock breaking. Bear theses usually arrive in a vacuum, giving the market time to sit with the discomfort. This one arrived between two announcements.

Tuesday's 4.49% decline suggests the discomfort has arrived late rather than not at all.

Nebius, Sovereign AI and the Bootcamp Funnel

Two further developments frame the next leg of the commercial expansion.

Palantir announced a strategic partnership with Nebius aimed at enhancing AI infrastructure capabilities, with sovereign AI customers identified as the target. Sovereign AI — governments and regulated institutions requiring AI systems that run within their own jurisdictions and under their own control — is a segment where Palantir's government heritage is a genuine competitive advantage rather than a legacy constraint.

The infrastructure partnership addresses the one thing Palantir does not own. Its platforms orchestrate models and data; they do not provide the compute underneath. Pairing with a dedicated AI infrastructure provider closes that gap without requiring the capital expenditure that hyperscalers are committing.

Separately, the company's sovereignty bootcamps drew nearly 200 organisations. That is the top of the commercial funnel, and it is the mechanism that produced 149% US commercial growth.

The bootcamp model is the operational core of the bull case. Rather than a traditional enterprise sales cycle running six to eighteen months, Palantir deploys into a customer environment for days, produces a working result against the customer's own data, and converts on evidence rather than on a pitch.

That compresses sales cycles, reduces customer acquisition cost, and explains both the margin structure and the retention figure. Customers who have already seen the product work on their data expand faster than customers sold on a roadmap.

The scalability question is whether the model survives volume. Bootcamps require Palantir engineers, and engineer capacity at 4,430 total employees is finite. Scaling commercial revenue tenfold on a bootcamp motion either requires proportional headcount growth — which compresses margins — or a partner channel that can run the motion independently.

The PwC alliance is the attempt to solve exactly that. It is the most important structural development in the business this year, and its success or failure will be visible in the commercial growth rate through 2027 rather than in the next quarter.

Burry's Thesis: Receivables, Deferred Revenue and Stock Compensation

The public short case is specific enough to be evaluated, which is unusual and makes it worth taking seriously.

Michael Burry, who first disclosed a bearish position in November of last year, posted on September 3 describing Palantir as "a consultant riding a bubble of AI FOMO demand" and warning that the roughly $432 billion market capitalisation could eventually fall below $100 billion. He added that even if the AI spending boom lasts for years, "the fall will be just as epic, or more so."

In May he stated Palantir was worth "low double digits at best." Following the August earnings surge he said he wished he had been short nearly $1 trillion worth of the stock.

The substantive argument rests on three accounting observations: rising receivables, low deferred revenue and heavy stock-based compensation.

Each has a real interpretation. Rising accounts receivable relative to revenue can indicate that revenue is being recognised faster than cash is collected, which in a services-heavy model would be expected. Low deferred revenue is the more pointed observation — genuine software companies collect subscription fees in advance, creating a deferred revenue balance that provides visibility. A business that bills in arrears for work performed looks like consulting on the balance sheet regardless of what the income statement says.

Heavy stock-based compensation inflates reported free cash flow and transfers cost to shareholders through dilution.

The counterevidence is the margin structure. Consulting businesses do not print 62% operating margins or 63% free cash flow margins at scale, and 157% net dollar retention is not a characteristic of project-based work.

The most honest reading is that Palantir's accounting profile sits somewhere between pure software and pure services, and that where exactly it sits determines whether 51 times sales is defensible or absurd.

A market capitalisation below $100 billion would represent a decline of roughly 76% from $418.88 billion — approximately $41 per share.

Valuation: Fifty-One Times Forward Sales and a 149 Price-to-Earnings

The valuation math is where this stops being a debate about the business and becomes a debate about arithmetic.

At $174.33 and a $418.88 billion market capitalisation, against full-year revenue guidance of $8.15 billion, Palantir trades at approximately 51 times forward sales. The trailing price-to-earnings ratio stands at 148.99. Enterprise value to trailing EBITDA of $2.66 billion runs above 150 times.

Fifty-one times sales is not a normal software multiple. Best-in-class enterprise software companies with 30% growth trade at 10 to 15 times sales. Companies growing 90% with 60% margins have historically topped out in the 25 to 35 range.

The bull argument for the premium is that the growth rate and the margin profile are both extreme, and that the combination compounds. At 93% growth, revenue doubles roughly every fourteen months. Two years of that takes the revenue base to roughly $30 billion, at which point the current market capitalisation represents 14 times sales — a defensible number.

The bear argument is that the growth rate is mathematically certain to decelerate, that 93% is not repeatable off a larger base, and that the multiple compresses faster than revenue grows once it does.

Sell-side positioning captures the disagreement precisely. The average twelve-month target sits at $193.88, implying 11.2% upside, with a high estimate of $255 and a low of $80. Twenty-one analysts recommend buying and two recommend selling. A separate consensus reading puts the average near $194.61.

A published target range from $80 to $255 is a 3.2x spread on the same stock for the same date. That is not ordinary disagreement about a growth rate — it is fundamental disagreement about what kind of company this is.

Add the short thesis targeting sub-$100 billion, or roughly $41, and the full distribution of published views spans from $41 to $255. The current price sits in the lower-middle of that range.

The macro overlay compounds it. With the 10-year at 4.80% and a September 16 rate hike priced near 60%, the discount rate applied to cash flows arriving in 2030 has risen materially, and no stock on the US market is more exposed to that than one trading at 51 times sales.

The Register: ARK Selling, Karp Selling, Retail Buying

Ownership flows have become the dominant short-term price driver, and the direction is not encouraging.

ARK Invest cut its position, selling approximately $27.2 million of Palantir while buying $27.5 million of SpaceX. That disclosure was the identifiable catalyst for Tuesday's 4.49% decline.

The pattern is not new. ARK sold 39,233 shares across two funds worth $6.38 million immediately following the August earnings surge, trimming into strength. Earlier in the year, during an April drawdown, the same manager bought 85,485 shares worth roughly $11.15 million.

That is disciplined position management rather than a thesis change — buying weakness, selling strength — but the market reads each disclosure as a signal, which amplifies the price impact well beyond the dollar size of the trades.

Insider activity has run the same direction. Chief executive Alexander Karp sold $53.6 million of stock. Founder-level selling at those amounts is routine for companies with large equity compensation programmes and pre-arranged trading plans, but it removes one category of buyer from a market that has relied on narrative support.

Retail sentiment on at least one social trading platform dipped over the past month and turned bearish, which matters more for this stock than for most given how much of the float sits in retail hands.

Notably, President Trump's accounts purchased Palantir stock during June, and options activity in the name has been consistently heavy, with one session described as showing massive action.

The composite picture is a register where institutional holders are trimming, insiders are selling, and retail sentiment is cooling — against a stock that gained 51% in the prior month.

That is the configuration that produces 4.49% declines on $27 million disclosures. Crowded positioning with no marginal buyer means every reallocation moves price disproportionately.

Levels: $182.53 and $207.52 Above, $162.66 and $123 Below

Map the structure.

Resistance begins at $182.19, Tuesday's high, and $182.53, Friday's close and the level established by the PwC rally. Above that, $186.38 marks the August month-end close. The 52-week high at $207.52 is the next structural reference and sits 19% above the current price. The high published target of $255 is 46% above.

Support starts at $173.67, Tuesday's low. Beneath it, $162.66 is the post-earnings close from August 4 and the first meaningful shelf. Below that, the pre-earnings base at $123.06 represents the July close and the level from which the entire August rally launched — 29% below the current price.

The 52-week low at $106.37 is 39% down. The low published sell-side target of $80 is 54% down. The short thesis implying sub-$100 billion market capitalisation corresponds to roughly $41, or 76% down.

Percentage distances from $174.33: resistance at $182.53 is 4.7% up, $207.52 is 19.0% up. Support at $162.66 is 6.7% down, $123.06 is 29.4% down.

The immediate technical question is whether the stock holds above $173.67. Losing it opens a gap toward the mid-$160s with little structure in between, because the August rally moved through that zone quickly and left limited accumulated volume behind.

That thin volume profile cuts both ways. A rally back through $182.53 would face equally little resistance up to the $186 to $190 area.

Volume at 27.96 million against a 30.99 million average argues neither side has committed. Genuine breaks in this name have historically come on turnover well above average — the post-earnings session traded far heavier than this.

Implied moves priced into options around events have run above 10%, and the stock has delivered 29% single-session moves. Position sizing calibrated to a normal large-cap does not survive contact with this instrument.

Verdict: Everyone Agrees on the Business and Nobody Agrees on the Price

Palantir at $174.33 with a $418.88 billion market capitalisation has produced the strongest quarter in its history and the argument about it has not moved an inch, because the argument was never about the numbers. Second-quarter revenue of $1.935 billion grew 93% year over year with adjusted EPS of $0.41 against $0.34 expected, US commercial revenue up 149%, US government up 90%, and simultaneous all-time highs in operating margin at 62%, net dollar retention at 157% and free cash flow margin at 63%, with full-year guidance raised to $8.15 billion and a third-quarter guide of $2.16 billion implying an $8.64 billion exit run rate. Nobody disputes any of that. The dispute is entirely about what those numbers are worth, and at 51 times forward sales, a 148.99 trailing price-to-earnings and above 150 times trailing EBITDA, the published distribution of views runs from a $255 high target to an $80 low target to a public short thesis arguing the market capitalisation belongs below $100 billion — roughly $41 a share. That is a range spanning 6x from bottom to top on the same stock for the same date, and it is why a $27.2 million fund disclosure knocked $18.8 billion off the market value on Tuesday while the September 3 PwC alliance had added roughly the same amount two sessions earlier. The specific bear points — rising receivables, low deferred revenue, heavy stock-based compensation — are answered by the 62% operating margin and refuted by nothing else, which leaves the balance-sheet character of the revenue as the genuinely unresolved question and the November 9 report as the next test. The path: hold $173.67 and reclaim $182.53 and the thin volume profile above opens $186.38 and then the $207.52 fifty-two-week high, 4.7% to 19.0% of upside. Lose $173.67 and the same thinness works in reverse toward $162.66, with $123.06 — the level the entire August rally launched from — the structural floor 29.4% below. Bias is unresolved between $162.66 and $182.53, and the honest framing is that with a 4.80% ten-year and a 60% September hike probability, the discount rate is currently working against the most duration-exposed large-cap on the US market — and a stock that moved 29% in one session and 51% in one month is not an instrument to size like a normal position in either direction.

That's TradingNEWS