Palantir grew 93% and paid a 1.4% tax rate — only one of those is in the headlines

Valuation — this is the entire bear case, and it isn't a small one

At $163.24 the market pays ~$391B (on basic shares; ~$419B on diluted) for a company guiding to $8.15B of 2026 revenue. That is ~48× forward sales. Large-cap software generally lives at 10–20×. For a same-day comparison against the most direct AI beneficiary in the market: Nvidia trades at ~20× trailing and ~12× forward sales, against Palantir's ~64× trailing and ~48× on the FY26 guide.

The earnings multiple is where it gets concrete. Trailing P/E is 139.6× and forward is 86.5× — but both are computed on a tax rate the company itself says is not the long-run rate. Normalize FY26: the adjusted operating income guide of $4.89B, less roughly $1.05B of stock compensation and payroll tax, plus ~$0.6B of interest and other income, taxed at Palantir's own 23%, lands near $3.4B of normalized earnings — about 114× the basic market cap (~122× diluted).

What has to happen. At ~114×, reaching a 30× multiple — still a premium to almost anything — requires earnings to grow about 3.8×. At a 40% earnings CAGR that is four years; at 30%, a bit over five. So today's price embeds roughly half a decade of compounding at rates almost no company sustains, just to arrive at "expensive."

One concession the bears should make: the de-rating already partly happened. Palantir entered this print ~41% below its 52-week high and down ~32% year to date — the business compounded at 85% through the first half of 2026 while the stock fell. Even after a ~30% two-day move it sits ~21% below the $207.52 high. The multiple compression bears were waiting for arrived in H1; this print bought some of it back.

The growth — what is actually accelerating, and what isn't

Take the operating business on its own terms, because it has earned that. Revenue growth went +70% (Q4 2025) → +85% (Q1 2026) → +93% (Q2 2026). Acceleration at a $1B+ quarterly scale is rare enough to be the story on its own. GAAP operating margin went 27% → 47% in a year — and this is the point most bear write-ups skip, because it is not a non-GAAP artifact; it is the reported income statement. Adjusted free cash flow margin is 63%. Guidance has gone from +61% to +71% to +82% in six months.

The margin story deserves one more line, because it is the opposite of what the standard Palantir bear case assumes. Stock compensation grew 65.8% — faster than total costs (+39.4%) — so it was a drag on margin, not its source. The decisive proof: the GAAP operating margin expanded 2,029bp while the adjusted margin expanded only 1,545bp. GAAP outran non-GAAP. Whatever else is going on in this print, the margin is not a stock-comp trick.

Cash agrees. Operating cash flow of $1,216M is 1.14× net income, and days sales outstanding compressed from 78.6 to 70.0 even as revenue grew 18.5% sequentially — receivables grew 5.7% against 18.6% revenue growth. Contract liabilities were a $144M source of cash. Customers are paying earlier, not later, which is the single most direct evidence against a "revenue is being pulled forward on paper" reading.

Then the counterweight. Closed TCV grew 49% while revenue grew 93% — TCV-to-quarterly-revenue fell to 1.74× from 2.26×. And the composition is narrow: US commercial TCV was $2.132B of the $3.373B total; back it out and bookings everywhere else went $1.427B → $1.241B, about −13% YoY.

Caveat, applied honestly — and this is the claim our own verification pass cut down hardest. Q2 2025 was a record TCV quarter at +140%, so the comp is punishing. The ratio actually rose sequentially, from 1.48× in Q1 2026. And TCV is the softest of three backlog measures: total remaining deal value is $13.1B, +83% — close to revenue's +93% — while remaining performance obligations, the non-cancelable measure, stand at $4.9B, with 43% expected to book as revenue inside twelve months. TCV is one yellow flag worth watching next quarter. It is not evidence the growth is hollow, and an earlier draft of this report leaned on it far too hard.

What the release doesn't put in its bullet list is worth naming, though it is a smaller point than it first appears. Customer count (1,049, +24%) and total RDV ($13.1B, +83%) were both given on the earnings call rather than in the highlights, and RPO appears only in Note 3 of the 10-Q. Nothing is hidden. But the +24% customer number tells you something the +149% US commercial revenue number does not: this quarter's growth is overwhelmingly existing customers spending more, not new logos arriving. That is a better business in the short run and a narrower one in the long run.

The shape nobody else in AI has

Palantir spent $14.6M on property and equipment this quarter — 0.75% of revenue — and generated $1.22B of adjusted free cash flow. Amazon's trailing capex is ~$169B; Microsoft's was $115.9B for the year with another $329B of uncommenced leases sitting off the statement.

That difference is not a detail, it is the investment case. Palantir sells into the AI buildout without funding it. There is no depreciation wave arriving in 2027, no debt raised to cover a capex hole, no negative-free-cash-flow window to explain away. The earnings-quality questions that dominated the Amazon and Microsoft prints simply have no place to live on this balance sheet.

Which is precisely why they moved into the price — and into the tax line.

Risk — each isolated, do not blur

1. Valuation / multiple risk (dominant). ~114× normalized earnings, ~48× forward sales. A 30% de-rating requires nothing to go wrong operationally — that is not a hypothetical, it is what happened in H1 2026 while revenue accelerated.
2. Earnings quality — the tax line. A 1.42% effective rate against the company's own stated 23% long-term rate. When the shield exhausts, reported EPS growth decouples downward from operating growth, and the optics of that quarter will be ugly regardless of the business.
3. Concentration. US revenue is 81% of total; US commercial supplied all the bookings growth. Marketed as a global sovereign-AI wave, the segment data currently describes a US phenomenon.
4. Bookings vs. revenue. TCV growing at half of revenue growth. If revenue keeps outrunning bookings, growth converges downward to bookings — the only question is when.
5. Non-operating income. $91.8M of "other income," ~14× the year-ago figure, sitting inside the 55% net margin. The 10-Q settles what it is: realized and unrealized gains on equity securities plus FX — and specifically $66M of net unrealized gains on publicly-traded stakes, "the majority of which are subject to short-term restrictions on the ability to sell." Palantir's public-equity holdings went from $23M to $184M in six months. So a visible slice of the quarter's net income is a paper mark on shares the company could not sell today if it wanted to, and it reverses if those positions fall.
6. Insider selling. Karp sold 397,744 Class A shares on May 20, 2026 under a 10b5-1 plan, with sustained executive-suite selling for two years. Pre-scheduled 10b5-1 sales are not a signal by themselves — but the aggregate is large enough to name rather than omit.
7. Competition (ranked low, deliberately). Every hyperscaler sells an AI application layer. This print made that case weaker: 149% US commercial growth is not what share loss looks like.

Time horizon & position sizing (kept separate)

Horizon. Not a next-quarter setup. The next two quarters are guided to decelerate — +83% for Q3, and the FY guide implies roughly +72% for Q4 — and Palantir has beaten its own guide eight quarters running, so the near term is mostly a game of by-how-much.

It is tempting to wave the deceleration away as a hard comp. That doesn't survive checking, and the check is clean because sequential growth is comp-independent. Through the back half of 2025 revenue compounded +40.2% from Q2 to Q4; the 2026 guide compounds only +25.3% across the same stretch. Hold Q2 2026's own delivered sequential rate (+18.5%) and Q4 lands near $2.71B — a +93% year-over-year rate, identical to the quarter just posted. So the entire 93% → 72% step-down lives in the guidance, not in the base. Either management is sandbagging by roughly $290M of quarterly revenue, or it sees something the print doesn't show. Given eight straight beats the first is more likely — but it is a choice about guidance, not arithmetic forced by the comparison. The thesis resolves over 3–5 years on the question the valuation section poses: does ~$8B of revenue become ~$30B+ before the multiple normalizes? Under two years you are trading sentiment on the AI complex, not the business.

Sizing considerations (not a recommendation). Two things separate Palantir from the other names in this section. It is the only one that doesn't carry capex-cycle risk — no depreciation wave, no debt, no negative-FCF window — so it is genuine diversification against AMZN/MSFT/NVDA exposure rather than more of the same beta. Against that: at ~48× sales, returns are dominated by the multiple rather than by operations, and H1 2026 is the proof — the business accelerated and the stock fell 32%. A position sized for the business will behave like a position sized for the multiple.