3Q26: Faster Revenue, Shorter Contracts

Thesis status: Affirmed in operating substance, with no condition weakening. Contract cover shortened and cost of revenue rose further above both markers set at 2Q26, and no condition sets a threshold on either.

This is Memo Update No. 3 to the operational semantics primer, after Memo No. 1 on 1Q26 and Memo No. 2 on 2Q26. The primer asks whether operational orientation produces commercial economics that compound durably and not cyclically, and these memos read it over 36 to 48 months from April 2026. This memo covers the third-quarter results and guidance released on 2 November 2026, valued at the 9 November close.

Key Takeaways

1. Palantir Technologies ($PLTR) reported third-quarter revenue of $2.31B, up 96% year on year and 20% on the quarter, 7% above the guide midpoint. US Commercial revenue of $984M grew 148% and passed US Government for the first time. GAAP operating income reached $1.08B, and management raised FY2026 revenue guidance to a midpoint of $8.53B, up 91%.
2. Contract measures did not keep pace with revenue. Total contract value (TCV) of $2.92B grew 6%, which took trailing book-to-bill to 1.78 times from 2.08 times. Remaining deal value (RDV) rose 2% on the quarter, to 1.84 times LTM revenue from 2.13 times, and remaining performance obligations (RPO) were flat. Management stated that multiple eight-figure deals were signed for shorter terms than usual, gave no figure, and did not repeat the duration-weighted TCV growth, bookings per year of contract term, it gave at 1Q26 and 2Q26. Those figures imply the average duration of deals closed was about a third lower than a year before, the reverse of Memo No. 1's reading, which this memo corrects. Net dollar retention (NDR) of 165%, billings and the raised guide show no weakness in current demand, but none tests forward bookings, so the disclosure cannot separate shorter terms from slower bookings.
3. Cost of revenue rose to 16.6% of revenue, further above both the 13% and 15% markers Memo No. 2 set. On the quarter it took 1.23 points of margin against 1.00 of leverage across operating expenses, so GAAP operating margin slipped 0.23 points to 46.9%. Year on year the ratio still fell nearly a point, although third-party hosting added $135M of the $176M rise in cost of revenue. The 10-Q discloses a March 2026 amendment committing at least $5.6B to one cloud agreement through February 2036, with annual minimums of $268M to $979M, where the primer recorded a $1.76B total at the FY2025 year-end. Management did not address hosting on the call.
4. At the 9 November close of $169.20, the ten-year revenue growth required on the constant-assumption series these memos track reads 34.3%, from 34.4% at Memo No. 2, as the larger revenue base removed 2.47 points and the higher price added 2.37, with smaller inputs removing 0.08. With the company's 23% long-run tax rate and a 10.96% discount rate built from current market inputs, the requirement is 39.3%, the first reading on that rate, and on Memo No. 2's 10.53% rate it reads 38.1%, against the 38.3% published then. All of these assume growth falls to 3% after year 10, and a higher exit multiple or a slower fade would lower the taxed ten-year figure by about 4 to 11 points by moving growth past year 10. Revenue compounded at 34.9% from FY2019 to FY2025, and at 41.7% to the FY2026 guide. Palantir trades at 58.3 times LTM revenue, against medians of 16.0 times for high-growth software and 6.6 times for all ten peers.
5. Three conditions stand Affirmed and five Developing, with none Weakening. Customers from the prior period supplied about 75% of trailing revenue growth on NDR's definition (derived). The ceiling on indefinite-delivery, indefinite-quantity (IDIQ) government contracts went unquantified again, and cash, equivalents and US Treasuries passed $10B with no capital allocation framework for a third quarter. Trends turned down on International Commercial, up 1% on the quarter, and stayed down on stock-based compensation, whose dollars rose 85% year on year. Competitive position turned down tentatively: Microsoft described Rolls-Royce's maintenance and supply chain systems as built on its own Foundry, and no document reviewed names a workload moved from Palantir. Insider sales since April were chiefly to cover tax on vesting, while plans adopted in the third quarter allow sales of up to 8.7M more shares, 7.1M of them Karp's, and a purchase of up to 4.0M by a Thiel-affiliated entity.

Third-Quarter Results

Revenue beat the guide midpoint by $151M (Exhibit 2) and adjusted operating income by $133M, and adjusted expense of $887M ran about $19M above the $868M the guide implied. Management had guided a significant third-quarter ramp in that line, and it rose 20% on the quarter against 14% at 2Q26.

Net income rose 6% on the quarter against 19% for operating income because other income swung to an expense of $26M from income of $92M, about $0.05 of diluted EPS, as 2Q26's gains on equity holdings did not repeat and 3Q26 recorded net unrealised losses of $25M (Exhibit 3). The 10-Q names no issuers, and most of the holdings are subject to short-term restrictions on sale. Adjusted EPS does not exclude these marks. Management gave no EPS effect for them this quarter, having quantified the SpaceX gain at 2Q26.

Operating cash flow held flat on the quarter as receivables absorbed about $410M, which took days sales outstanding to about 75 from about 70, below the 78 of a year earlier (derived). One unnamed customer accounts for 27% of receivables, little changed from 25% at the year-end, while no customer exceeds 10% of revenue.

Outlook

Management raised FY2026 revenue guidance by $378M at the midpoint, $151M of it the third-quarter beat and $227M the fourth quarter (Exhibit 4). The fourth-quarter guide of $2.649B to $2.653B implies growth of 88%, and 15% on the quarter. The third-quarter guide implied 83%, against a result of 96%. Adjusted operating income of $1.640B to $1.644B implies a 61.9% margin and adjusted expense of about $1.01B, up about 14% on the quarter (derived). The US Commercial floor implies at least $1,216M in the fourth quarter, up 140% year on year and 24% on the quarter (derived). The guide reaches one quarter ahead, and management gave no figure for cost of revenue, stock-based compensation or bookings.

The Contract Book

Revenue growth has run ahead of TCV growth for three quarters, and the gap widened to 90 points in the third (Exhibit 5). TCV is lumpy and the year-ago comparison was strong, so one quarter overstates the change. On trailing measures the crossing came only now: LTM TCV growth of 56% fell below LTM revenue growth of 87% (both derived) for the first time in at least a year, after 92% against 79% at 2Q26. Trailing book-to-bill fell for a third quarter, and RDV fell again relative to revenue (Exhibit 6).

US Commercial explains the sequential fall. Its TCV dropped 31% from $2.13B at 2Q26, which had risen 81% on the quarter, and $1.46B still exceeds every quarter shown except 2Q26. Over twelve months US Commercial TCV grew 62%, against 117% at 2Q26. TCV outside US Commercial rose on the quarter and stood about 1% above a year earlier, which hides a split. On commercial TCV of $1.65B, up 17%, government TCV of about $1.27B was about 6% below a year earlier and 22% above 2Q26 (derived), in a quarter when government revenue grew 81%. In US Commercial, RDV rose $200M less than TCV net of revenue would imply, against about $50M at 2Q26. Palantir does not bridge RDV, and the gap could reflect terminations, lapsed options, contracts replaced on conversion or exclusions for customers' financial condition, which the release allows for.

Long-term RPO fell as short-term RPO rose, taking the share due within twelve months to 44%, from 43% at 2Q26 and level with a year earlier (Exhibit 7). RPO is partial and mostly commercial, and it excludes contracts with original terms of twelve months or less, so any deals signed below that line add to RDV and never reach RPO, and part of the flat reading may come from that definition.

Management stated that multiple eight-figure deals in the quarter were "shorter in duration than usual as customers rapidly adopted initial Sovereign AI capabilities", and that it expects both duration and size to increase over time. Karp described "shorter duration of the contract, but then an ability to expand that contract dramatically". Deal counts are compatible with that account without establishing it: deals of at least $10M fell to 64 from 73, though they stood at 47 at 1Q26, while deals of at least $5M rose to 107 from 98. The size of the effect is undisclosed. The release still defines annual contract value (ACV) and the dollar-weighted duration basis, and management quoted individual deals in ACV, but it gave no aggregate figure on either measure.

The earlier disclosures frame the size of the question (Exhibit 8). Duration-weighted growth at 2Q26 implies the average duration of deals closed was about a third lower than a year before, in total and in US Commercial, and the 1Q26 figures read the same, so the fall is not a third-quarter effect alone. The measure cannot separate shorter contracts from fewer large long deals. In both first-half quarters duration-weighted bookings grew faster than revenue. Closing the gap at 3Q26 would need average duration down 46%, and 55% in US Commercial, and a fall on the first half's scale would leave annualised bookings growing well below revenue's 96%. The disclosure that would settle it was not given.

Correction to Memo No. 1. Memo No. 1 read duration-weighted growth of 135% as customers "committing to longer contracts at a faster rate" than short-cycle expansions, a pattern it called more consistent with structural adoption than cyclical buying. Management stated 61% and 135% on the 1Q26 call and did not describe contracts as lengthening. On the definition in the 2Q26 and 3Q26 releases, growth above reported TCV means the average duration of deals fell. The figure does not support a reading of longer contracts, and the structural-adoption inference drawn from it is withdrawn.

Two readings remain open, and the quarter cannot separate them. In the first, customers are buying shorter initial terms and then expanding, so TCV understates demand. In the second, annualised bookings have slowed behind revenue, so current growth is drawing on a book that is filling more slowly. NDR, billings and the raised guide do not decide between them because each reads the present: NDR looks back twelve months, billings are mostly revenue and rose more slowly than it on the quarter, and the guide covers one more quarter. On either reading more of future revenue rests on renewals, expansions and new customers over the next few years, and no condition sets a threshold on contract cover.

Who the Growth Comes From

Net dollar retention divides the trailing revenue of customers counted in the prior twelve months by what those customers paid in that prior period, so its definition splits growth in two (Exhibit 9). Customers from the prior period supplied about 75% of LTM growth, within a 72% to 75% range for three quarters. They are prior-period customers and not purely expansion, because the cohort holds customers acquired during the prior period, whose part-year revenue sits in the denominator. Management stated at 1Q26 that the rise in NDR "was driven both by expansions at existing customers and new customers acquired in Q1 of last year".

The 10-Q asks the same question from a fixed date, splitting the year-on-year increase by customers on the book at 31 December 2025 (Exhibit 10). Customers new in 2026 supplied 21% of the increase, from 10%, and $124M of the $222M sequential rise in commercial revenue. The cohort grows through the year by construction, so the rise alone does not show acquisition accelerating. Net additions of 124 commercial customers this year imply at most about $1.71M of revenue a quarter per new customer, against about $1.23M across the existing base (derived), since gross additions exceed net. The ceiling sits above the existing-base average, which keeps open the possibility Memo No. 2 raised, that new customers are entering larger, without showing it.

Management's examples would fit, if these are new accounts, which the call does not say. An industrial materials company went from a pilot of nearly $3M to $29M of ACV, a global athletic company from $1.6M to about $21M, a global home appliance company from $1M to $15M and an apparel company from $2.5M to $11.5M, each starting in the first half of 2026. New accounts would enter NDR's cohort next year with pilot-sized revenue in the denominator, and their conversion would then read as retention.

Cost of Revenue and Hosting

This section is the standing overlay Memo No. 2 opened, which holds no status and no thresholds. Memo No. 2 set two markers for the ratio, nearer 13% pointing to a customer-specific accommodation and nearer 15% to a repeatable hosting model. The ratio rose further above both, giving back just over half of its gain from 1Q25 to 1Q26 (Exhibit 11).

Of the $86.4M sequential increase in cost of revenue, $7.0M is stock-based compensation and $79.4M is delivery cost (derived). Excluding stock-based compensation the ratio rose to 14.9% from 13.7%, and adjusted gross margin fell to 85.1% from 86.3% (derived). Third-party cloud hosting is the line the 10-Q names, on a year-on-year basis only (Exhibit 12), and it also added $21M to research and development. Management neither repeated nor updated the 2Q26 explanation of cloud hosting taken on for one government customer.

The markers do not settle the reading. The level points toward a repeatable model. The segment figures keep a single scaling customer possible, because the 10-Q allocates hosting directly to the account it serves, and contribution margin fell in Government to 69.6% in the third quarter from 72.0% in the first half, while it rose in Commercial to 76.9% from 75.5% (derived). Note 7 adds scale without attribution: annual minimums of $268M to $979M compare with LTM cost of revenue of $1.11B, and the note names no counterparty and ties the agreement to no customer.

On the quarter the rise in cost of revenue outweighed leverage across all three operating expense lines (Exhibit 13). Incremental GAAP operating margin fell to 46%, from 52% in the second quarter and 79% in the first (derived), and the gap between adjusted and GAAP operating margin held at about 15 points. One quarter does not establish a ceiling on margin, and Exhibit 18 shows what a lower terminal margin does to the requirement.

What the Price Implies

This section characterises the financial conditions the current enterprise value implies. It is not a price target, a fair-value estimate, or a view on whether the shares are mispriced.

At the 9 November close of $169.20, on 2,574M diluted weighted-average shares and $10.4B of net cash, the market values Palantir's operations at $425.1B (Exhibit 14). One instrument is built two ways. In both, revenue compounds at a constant rate from the LTM base, margin moves in equal steps from the LTM GAAP margin to 45% at year 10, stock-based compensation is expensed, capital expenditure and working capital are left out, and growth then falls to 3% in perpetuity. The record series applies no tax and holds the frozen 10.53% rate, so each reading compares with the last. The taxed construction applies 23% to every flow and, from this memo, the native 10.96% rate. The LTM margin of 45.6% now sits above the 45% input, so the margin path descends for the first time, and the input stays at 45% as a constant assumption.

The record series fell 0.18 points (Exhibit 15). The larger revenue base and the higher price netted 0.10 of that, and margin, net cash and shares supplied the other 0.08. The order of the steps barely matters: taken first, the base removes 2.44 points, and taken last, the price adds 2.35. This is arithmetic and not a read on the shares. Year 10 now ends a quarter later, so year-10 revenue on the series rose 16.9%, or about 8.6% at the same end date, and at a flat price an FY2026 base at the guide midpoint would remove about 2.3 more points.

The close has ranged widely since 30 June. It stood at $116.67 on 30 June and $209.05 on 9 October, then at $178.34 on 2 November, the last close before the release, $160.96 on 6 November, 9.7% lower over four sessions, and $169.20 on 9 November. Exhibit 16 re-solves every close on this memo's inputs, so it varies price alone and does not show what the requirement read on each date. This memo attributes none of these moves to a cause.

Revenue compounded at 34.9% from FY2019 to FY2025, and at 41.7% to the FY2026 guide and 40.3% to LTM, from a base of $743M over six to seven years. Both requirements ask for ten years from $7.3B. Year-10 revenue of $200.0B on the taxed path is 2.8 times Oracle's LTM revenue today and 58% of Microsoft's, before any growth at either.

Palantir trades above all ten peers on revenue, with CrowdStrike next at 41.2 times, and it grew faster and converted more revenue to free cash flow than any of them (Exhibit 17). The gap narrows on free cash flow, where 104.9 times sits below CrowdStrike's 145.7 times and above Snowflake's 78.9 times, though both spend a far larger share of revenue on stock-based compensation, which free cash flow does not deduct. The set is bimodal, so the group medians say more than the all-peer median.

Terminal construction moves the taxed requirement more than the other cases tested (Exhibit 18). Growth falling at once from 39.3% to 3% after year 10 implies an exit of 9.96 times EBIT, and terminal value is 74% of enterprise value. Exits at today's peer medians of enterprise value to GAAP operating income, which Exhibit 17 does not show, of 14.0 times, for government and defence IT, and 19.8 times, for the seven peers with a meaningful multiple, read 35.38% and 31.44%, and a ten-year fade reads 28.23%. These lower the ten-year rate by moving growth past year 10: that fade still reaches $368B of revenue by year 20, against $269B on the base path. The requirement is conditional on this construction. A cut to a 40% terminal margin adds 1.65 points on the taxed construction and 1.59 on the record series, near the 160 bps Memo No. 2 published on that series. A rise to 50% removes 1.46 and 1.41.

The Conditions

No status changed from Memo No. 2. Trends eased to unchanged on net dollar retention and US Commercial growth, turned down on competitive position and International Commercial, and stayed down on stock-based compensation (Exhibit 19).

C1: Net dollar retention above 130%

🟢 Affirmed → Retention of 165% clears the threshold by 35 points, with the mechanism read only in dollars.

The 10-Q attributes growth at existing customers to "increased adoption of our products and services within their organizations", and average LTM revenue from the top twenty customers rose 76% to $146M. Revenue outside the top twenty rose faster, at about 95%, which took the top twenty's share to about 40.1% from 42.6% (derived). The trend moves to unchanged from improved because the headline rose while the prior-period share of growth held.

C2: US Commercial growth above 60%

🟢 Affirmed → Growth of 148% stands 88 points above the threshold, while the forward measures slowed behind it.

Revenue per customer reached $1.43M in the quarter, from $0.75M a year earlier, on customer growth of 29%. The trend moves to unchanged from improved because the headline held as US Commercial TCV growth fell to 11% from 153% and US Commercial RDV growth on the quarter fell to 4% from 27%.

C3: Government ceiling conversion

🟡 Developing → US Government revenue grew 96%, and the ceiling the condition names was not quantified again.

The last figure on hand is FY2025's $12.3B, which Memo No. 2 mislabelled as a 1Q26 figure. Palantir does not disclose task orders. The nearest proxy, RDV excluding US Commercial, which also holds the international segments, was little changed in dollars and fell to 1.55 times its LTM revenue from 1.76 times (derived). The Army awarded a prime agreement to produce eight TITAN systems, with no value given, and Maven exceeded 100,000 users and is now, on management's account, available to all services. Revenue is strong and the forward proxy weakened, so the evidence is mixed and the trend holds, while the condition's own measure stays unreadable.

C4: Competitive position in operational workloads

🟡 Developing ↓, tentatively. No document reviewed names a workload moved from Palantir, and Microsoft's call meets the primer's product test at announcement level. The trend is tentative because Memo No. 2 read none of these disclosures, so the comparison is with Memo No. 1 and the primer.

The primer named three forms of evidence: customer migrations, win rates in disclosed competitive situations, and product announcements extending Fabric or SageMaker into Ontology territory. The first two did not appear. On the third, Microsoft is bringing Fabric closer to its own Foundry, which is unrelated to Palantir's product of the same name, and states that over 19,000 customers use both "as enterprises connect and operate agents over real-time operational, analytical, and unstructured data". Nadella described Rolls-Royce's maintenance and supply chain systems as "all built on Foundry agentic systems" (Exhibit 20). Microsoft also named sovereignty over an organisation's own learning one of three priorities, the axis on which Palantir began marketing itself at 2Q26, and Amazon made a related data-control claim for Bedrock.

Two readings fit. The competitive gap may be narrowing, or hyperscalers may now sell against Palantir's claim without having taken a workload from it. A sovereignty claim that Microsoft also makes is harder to hold as a point of difference, which bears on the question Memo No. 2 left open without settling it. Overlap is unsized, and nothing shows any named account was a Palantir account. Against all of this stand US Commercial growth of 148% and NDR of 165%.

Palantir's own competitive frame moved to the labs. Sankar set "the model makers" against diffusion, cited page 209 of Anthropic's S-1 on AIG and stated "We are the diffusion layer, not just of choice, but of necessity", and Karp described "a symbiotic relationship" with Anthropic. No hyperscaler was named on the call. Memo No. 2 sourced the Nvidia partnership to press, though Karp named it on the 2Q26 call. Page 5 of this quarter's Business Update shows Nvidia training open models on supply chain data in the Ontology, with no value given, and the 10-Q does not mention it.

C5: International Commercial beyond the UK

🟡 Developing ↓ International Commercial revenue grew 21% year on year and 1% on the quarter, after 2% at 2Q26, and the leg beyond the United Kingdom stays unreadable.

Note 12 reports only United States and Rest of world, with no other country reaching 10% of revenue. Neither of Karp's two references to Europe gave a commercial read: one, in his opening remarks, was "we should kind of avoid talking about it", and the other concerned allies. The Accenture partnership was not mentioned. The 4Q25 base of about $171M means $188M next quarter would equal 10% growth, the rate below which, across multiple years, the primer reads the gap as permanent. International Government grew 31% and 6% on the quarter, so the weakness is commercial.

C6: Stock-based compensation restraint

🟡 Developing ↓ The ratio rose to 13.8% in a quarter when it fell in each of the prior two years, and dollars rose 85%.

On the rounded dollars shown, stock-based compensation rose 20.4% on the quarter against revenue's 19.5% (Exhibit 21), so the larger denominator took 2.69 points off the ratio against 2.79 added by the numerator. The ratio stays below 4Q25's 13.9%, which keeps the status at Developing. Memo No. 2 read the guided third-quarter ramp as a guide to stock-based compensation. Management guided it in adjusted expense, a line that excludes stock-based compensation, so the guide did not address the ratio and Memo No. 2 should not have read it that way. The 10-Q attributes the 85% year-on-year rise in dollars to new grants since 30 September 2025 and gives no sequential explanation, and RSUs granted in 2026 averaged $130.66 against $27.74 on the opening balance, which lifts dollars by construction.

Unrecognised expense stands at about $2.0B, over weighted periods of four to eight years. Sankar, Glazer and Taylor received 987,246 shares at no cost on 28 October, worth $182M to $194M at the 30 October and 23 October closes (derived), which post-date the quarter and fall outside that figure. The forms give no vesting terms. If the awards vest over time they add to expense across future years, and if they were fully vested at grant they land in the fourth quarter. Diluted shares rose 0.11% year on year and basic shares 1.2%.

C7: Capital allocation of the cash position

🟡 Developing → Cash, equivalents and US Treasuries reached $10.4B, a third quarter with no framework.

Part II Item 2 reports no repurchases in the quarter and $1.5M over nine months, against $22.6M from option exercises, and capital allocation was not raised on the call. None of the five items Memo No. 2 listed was supplied, from the purpose of the balance to the conditions for repurchase. Karp's one remark on value was that the market value of the commercial business "probably justifies our current valuation", which states no policy.

Correction to Memos No. 1 and No. 2. Memo No. 1 set Karp's "somewhat undervalued" beside buybacks as the first directional movement at the rhetorical layer, one that might compound into policy, and Memo No. 2 carried it as a rhetorical signal on capital allocation that did not compound. On the 1Q26 call he made the remark in an answer on recruiting, ending "so it's a great place to go". Neither memo should have read it as bearing on capital allocation.

C8: Key-person and founder voting events

🟢 Affirmed → No leadership or voting change, and insider selling since April was chiefly to cover tax on vesting.

This is the first full read since 1Q26, from Part II Item 5 of the 10-Q and the Form 4s from 1 April (Exhibit 22). Karp's net increase of 1.06M shares is a vesting effect, as 1.95M RSUs vested against 890,092 shares sold. No form shows a Thiel sale in the period, against the 2.0M on 2 March recorded in Memo No. 1.

The plans widen the picture. Sale plans adopted in the quarter allow up to 8.73M shares, about $1.48B at the mark. Karp's 7.08M is 12% of his reported direct Class A and Class B shares, excluding options and RSUs. Neither his plan nor the Thiel-affiliated purchase plan shows a transaction by 4 November, which this early in their lives says little. The Founders can still control up to 49.999999% of the vote.

Thesis Standing

Operational orientation is producing the economics the primer described, at a scale it did not assume. The open question has moved from whether those economics exist to how long the contract book can show them. Retention, revenue and the guide all read the present, and the measures that reach further out shortened in the same quarter, without the figure that would size the effect.

If shorter terms are a first step toward larger contracts, as management expects, duration and TCV should recover in coming prints, and durability gets tested sooner because contracts come up for renewal sooner. If they are not, the book keeps filling more slowly than revenue grows, and the first sign would be RDV cover falling further while NDR holds, since NDR looks back twelve months.

On this construction the price depends on years the book cannot show. RDV covers 1.84 years of LTM revenue, and terminal value alone is 74% of enterprise value on the taxed construction. Cost of revenue is the second live exposure, moving through the margin input, and the third quarter does not show whether it reflects one customer scaling or a changed delivery model.

What to Watch

The 4Q26 print tests the contract book against the $4.26B of TCV in 4Q25, and shows whether the duration-weighted figure returns (Exhibit 23).

Sources: Palantir Form 8-K with Exhibit 99.1, Q3 2026 Business Update and earnings call, 2 November 2026. Palantir Form 10-Q for the quarter ended 30 September 2026. Palantir Forms 4 from 1 April to 4 November 2026. Palantir 1Q26 and 2Q26 earnings calls, and 2Q26 release, Business Update and Form 10-Q. Microsoft Form 10-Q for the quarter ended 30 September 2026 and earnings call, 28 October 2026. Amazon Form 10-Q and earnings call for the third quarter of 2026. Snowflake Form 10-Q for the quarter ended 31 July 2026. Peer figures from each company's latest filing or release as logged in the Palantir reverse DCF model. The primer, Memo No. 1 and Memo No. 2. Share prices of $144.45 at the 4 August close, $178.34, $160.96 and $169.20 at the 2, 6 and 9 November Nasdaq closes, and the other closes from 30 June logged in the model for Exhibit 16 and C6, with peer prices at 9 November. Requirements solved on the construction stated in the valuation section, with reference to the Palantir reverse DCF model v3.4, on 2,573.9M diluted weighted-average shares, net cash of $10,376M, a 10-year yield of 4.91% at the 9 November close, Damodaran's implied equity risk premium of 4.28% at 1 August and a beta of 1.4143. Adjusted operating income, adjusted expense, adjusted gross margin and adjusted free cash flow are non-GAAP measures as Palantir defines them. TCV, RDV, RPO and NDR are as Palantir defines them, and RDV and TCV assume all options are exercised._