Last week, I mapped out the every dollar DoW had to spend in FY2026, traced from the bill that provided it down to individual programs. Click around the full interactive chart here.

On October 13, Anduril opened a tender offer letting current and former employees in 16 countries sell up to $1.3 billion of stock at $94.80 a share. The price valued the company at roughly $90 billion.

So for this week, I put together a financial model on Anduril to answer the question: what has to be true for the market to pay that price?

The full model has ten tabs covering Anduril’s product catalog, federal obligations, defense-industry comps and the investor-return math, with every input sourced and graded hard, soft or assumption. [Make your own copy here] and tell me what I got wrong.

Since its Series B in 2019, Anduril's valuation has gone from $1 billion to $90 billion while its revenue grew roughly fortyfold. Palmer Luckey has said he started Anduril to save taxpayers hundreds of billions of dollars while making tens of billions a year, and at no point along the way have investors priced Anduril like a defense contractor. At Northrop's valuation multiple (~1.9x), Anduril would need ~$44B in revenue to justify its price, about ten times what it expects to book this year.

So a $90 billion price tag today is a bet that Anduril never becomes a legacy prime, even as it's beginning to show signs of one, and that it stays a hypergrowth, software-margin company at a scale no defense hardware maker has ever reached.

Setting Anduril's $90 billion against Northrop's $73 billion compares the value of their shares. But if you were to buy either, the whole business inherits its debt and keeps its cash, so I compared enterprise value instead.

Northrop carried about $12.9 billion of net debt on its June balance sheet, which lifts its enterprise value to roughly $85 billion. Anduril obviously doesn't publish a balance sheet, but after raising $5 billion in May and spending heavily on two factories, I estimated it holds about $4 billion of net cash, which puts its enterprise value near $86 billion or roughly equivalent.

Northrop ($NOC ( ▼ 0.05% )) reports third-quarter earnings today (October 20), and its stock has already climbed back from a 52-week low set earlier this month. The company expects $44 billion in revenue this year and sits on a record $105 billion backlog. Anduril, on the other hand, expects $4.3 billion in revenue and a $1.2 billion operating loss.

The legacy primes trade around two times sales. Kratos and AeroVironment, the companies whose drones, munitions and counter-drone systems are closest to Anduril's catalog, trade at about three. At the far end sit Palantir and SpaceX, both above 55 times sales and growing 80 to 90 percent a year with margins of 45 to 60 percent. Anduril's tender lands about a third of the way from the drone makers to Palantir, so the market is paying as if a third of Anduril were a software company and the rest a very good hardware business.

Among the contracts I could trace to dollar values, the work identifiable as Lattice comes to about $50 million, roughly 1 percent of this year's projected revenue. Anduril builds Lattice into its drones, towers and interceptors, so a line-item count understates it, but it also means the premium rests on margins we can't actually trace yet.

For comparison, NG splits $42 billion in sales across four businesses of roughly equal size: Aeronautics ($13.0B), home to the B-21 bomber, E-2D and F-35 fuselage work; Mission Systems ($12.5B), which builds radars, electronic warfare and command-and-control systems; Space Systems ($10.8B), satellites and classified space programs; and Defense Systems ($8.0B), the Sentinel ICBM, the IBCS missile-defense network, GMLRS and ammunition.

Much of that is sole-source work measured in decades and the market pays less than two times sales for it because growth is slow but there is little doubt the revenue will show up.

This is where Anduril is beginning to bark like a prime. In March the Army signed an enterprise agreement worth up to $20 billion. In June the Air Force picked Anduril's Fury drone for production. In October the Navy awarded up to $2.9 billion for Virginia-class submarine components, and the Army up to $1.8 billion years to expand its next-generation command-and-control network.

But you still have to read the fine print. The $20 billion is a ceiling over 10 years, and the agreement consolidates more than 120 existing purchasing actions, so much of it covers revenue Anduril already books. The submarine money pays out only as Arsenal-2, which won't open until 2030, delivers parts.

The NGC2 award only commits $162.8 million up front and is on a 5 year timeline. Anduril has said Fury can't enter production until Congress passes a fiscal 2027 budget. And the Barracuda-500M missile deal guarantees at least 3,000 missiles without a disclosed price.

For calendar 2025, you can trace about 62 percent of the $2.2 billion Anduril reported* The missing $830 million most likely comes from international customers such as Australia's Ghost Shark program, classified work, and revenue recognized on money obligated in earlier years. Calendar year 2026 is already a record with more than two months of data still to post, but a continuing resolution and limits on new starts put the back half of the $4.3 billion projection at risk.

*Note: In Anduril’s Series H Announcement, the company reprinted their investor letter which stated the they more than doubled revenue in 2025, reaching $2.2B.

The fair objection to everything above is that Anduril isn't a prime and shouldn't be valued like one. That's right, so instead of asking whether Anduril deserves Northrop's multiple, I asked how Anduril's fundamentals stack up against Northrop's across a range of multiples.

Holding Northrop at its current multiple and growing it 6 percent a year, puts its enterprise value near $108 billion by 2030. I compared that against two Anduril revenue forecasts: a top-down one that grows its existing business 20 percent a year and layers on its big production programs, and a bottom-up one built from the contract catalog.

At a prime's multiple, Anduril would need about $58 billion of revenue in 2030 to match Northrop. At the three-times multiple drone makers earn, it would need about $33 billion. At 10 times sales, the requirement falls to about $11 billion, and at 20 times, where Anduril trades today, it already has the projected revenue.

Multiples fall as growth slows, and that's the hard part. Kratos is growing about 32 percent this year and trades at about 3.4 times sales. On a strong but plausible path, Anduril will be growing about 30 percent by 2030. To command 10 times sales then, it would need to be either still growing 50 percent or more on a base above $10 billion, or earning margins above 30 percent, the kind software companies earn.

It's tempting to say none of this matters. Anduril has the capital, the talent and the access, and its backers won't let it fail even if public reporting has stated their products repeatedly come up short. But whether Anduril succeeds and whether $90 billion is a good price are different questions, and the answer to the second depends on when you bought.

Founders Fund led Anduril's seed round in 2017 at a valuation of about $88 million. At $90 billion that stake is up roughly a thousandfold before dilution so it can hardly lose. The investors who came in at $61 billion in May are up about one and a half times.

For late-stage investors to triple their money over about five years, working backward from $90 billion and assuming the market pays 10 times sales at exit, Anduril would need about $31 billion of revenue by 2031, growing about 49 percent a year from here. The top-down forecast reaches about $12 billion by then, which would leave a $90 billion buyer roughly breaking even. A company that keeps growing fast but slows gradually, from 70 percent next year to 25 percent in 2031, gets to about $25 billion and returns roughly 2.4 times.

So maybe $90 billion isn't absurd. It assumes Anduril keeps executing like an exceptional company for five more years. But that's a lot to ask of a company that has never produced aircraft or submarine components at scale, with a skilled workforce that by many accounts doesn't exist anymore.

The strongest argument against this analysis is that Lattice lives inside the hardware. Anduril's software runs its drones, towers and interceptors the way Tesla's runs its cars, so revenue that looks like hardware may carry software-like margins. A related view holds that the premium reflects growth, not software: Anduril is growing about 95 percent this year, and Palantir's 56-times multiple rests as much on 82 percent growth and 60 percent margins as on being a software company.

Tax payer saving are another bonus Anduril's advocates often cite. The unit prices are far below incumbents': a Fury at an estimated $24 million against an F-35A at about $82 million. But a unit-cost comparison only counts as savings when one system replaces another.

A drone wingman does not replace a crewed fighter, and a Barracuda with a 100-lb warhead definitely doesn’t replace a Tomahawk. However, cheaper per unit, in categories where the government is buying more units, can mean more total spending, not less.

In April 2017, Tesla passed Ford in market value on a fraction of Ford's sales, and went on to earn it. In November 2021, Rivian did the same thing, then lost most of its value and hasn’t recovered yet.

Anduril passing Northrop is the same kind of moment, with three things worth watching.

Do ceilings turn into obligations: production money for Fury once the fiscal 2027 budget passes, Barracuda deliveries next year, and command-and-control task orders beyond the initial $162.8 million.

The next private round. A round at or above $94.80 a share would say investors still believe the story; a flat or lower one would be the first crack.

Will revenue keep pace. A strong path runs through roughly $7 billion next year and $11 billion in 2028, and falling behind early is how a promised tripling quietly becomes a break-even.