Anduril Industries

IWhat it does

Anduril is a US defense contractor that was founded in 2017²³ and is still privately held³³. It builds machines that fly, swim and sit on the ground, plus the software that connects them. The idea at the start was simple. The US government had spent decades buying a small number of very expensive systems from a small number of large contractors. Anduril set out to build cheaper, more numerous machines that are guided by software, and to fund the work with its own and investors’ money before it had a contract. In the words of its chief executive, defense was not a category that attracted much venture investment when it began¹.

The software is called Lattice. It takes data from sensors and weapons and lets operators and machines act on a shared picture. The US Army has used it as the base for its next-generation command and control system, which is meant to replace an older system that manages fires, logistics and airspace⁹. The hardware list is long. Anduril’s own site describes the Fury autonomous air vehicle, which press coverage calls a collaborative combat aircraft¹²¹³²⁵, the Roadrunner reusable air vehicle for air defense²⁶, the Barracuda family of air-breathing autonomous air vehicles²⁷, the Altius autonomous air vehicle launched from ground, air or sea²⁸, the Ghost Shark autonomous underwater vehicle²⁹ and the Sentry autonomous surveillance platform³⁰, among others. I treat that list as a map of the product line, not a verified catalogue. A product on the list is not proof of revenue.

The company has been building its own factories. Arsenal-1 is a 5 million square foot site next to Rickenbacker airport near Columbus, Ohio. Anduril says it has committed about $910 million to it, and the first complete Fury aircraft built there rolled out on 27 July 2026¹². Defense One reported in March 2026 that the plant was due to begin production months ahead of the July 2026 date Anduril had given in 2025⁵. A second site, Arsenal-2, was announced on 6 October 2026. It is a planned shipyard in Maryland for Virginia-class submarine components, funded by $3.7 billion of Anduril capital and a Navy contract worth up to $2.9 billion, with first operations expected in 2030⁸¹¹.

IIHow it makes money

Anduril sells almost entirely to governments. The largest customer is the US Department of Defense (now styled the Department of War in some official documents) and its services: the Army, Air Force, Navy, Marine Corps and Space Force. Other customers named in press coverage include the Dutch, UK and Australian defence ministries, NATO and Poland². Revenue comes from several kinds of contract. Some pay for development of a system. Some are production orders for hardware. Some are software subscriptions or support. I could not find a public split between these, and the company does not publish audited accounts, so I cannot say what share is hardware and what share is software.

The Army enterprise contract is the clearest structural change. On 13 March 2026 the Army announced a 10-year deal with a ceiling of up to $20 billion, made of a five-year base period and a five-year option. It rolls more than 120 separate buying actions into one framework and sets pre-agreed prices and volume discounts⁴. The Army states plainly that $20 billion is the maximum possible value, not money it has committed, and that the deal does not replace competition for future programs⁴. For comparison, the Army signed a similar $10 billion ceiling deal with Palantir in August 2025, according to press coverage of the Anduril award⁶. A ceiling is not revenue, and I do not count it as backlog.

Other awards are better described by what is firm. On 30 September 2026 the Army awarded an initial $162.8 million to move its next-generation command and control software into fielding in the Indo-Pacific, with options that can raise the contract to $1.8 billion over five years⁹. The Air Force moved Anduril’s Fury aircraft, and General Atomics’ competing design, from prototype into production on 17 June 2026, and says future lots of this first batch of drone wingmen are to be bought only from those two firms¹². The Navy contract of up to $2.9 billion pays against demonstrated production outcomes, so Anduril carries most of the delivery risk¹¹.

On 13 May 2026 the Pentagon announced framework agreements with Anduril, CoAspire, Leidos and Zone 5 under a low-cost containerized munitions programme. It could buy more than 10,000 missiles over three years from 2027, and the agreements set terms for later fixed-price production contracts¹⁶. Several outlets described these as hypersonic missiles. Defense News describes the programme as cruise-type containerized munitions and says the hypersonic Blackbeard deal is a separate one with another company, Castelion¹⁶. I follow Defense News. Press reporting said Anduril planned capacity of 1,000 missiles a year, with the first batch in early 2027, but I did not open the source for that and leave it out of the numbers. The practical point for an owner is that a framework agreement fixes terms, not volumes: each later production order has to be won and funded¹⁶.

Contract and programme values, and what each number actually is

IIIThe market and the competition

Anduril sells into the defense budget, which is large but slow. The US reportedly plans more than $26 billion in the fiscal 2027 budget for multi-year munitions contracts alone, according to testimony quoted by Defense News¹⁶. Venture funding of defense start-ups also rose: the Financial Times, as relayed by TechCrunch, counted over $12 billion in the first half of 2026, more than all of 2025². That is good for demand and for access to capital. It also means more rivals are funded to chase the same programmes.

Competition depends on the product. In collaborative combat aircraft the rival is General Atomics, and the Air Force chose both¹². Northrop Grumman had also competed for the first aircraft, per Defense One⁵. In software and command and control, Palantir is both a partner and a competitor. Defense News names Palantir, Raft and Striveworks as companies that built the Army’s common data baseline alongside Anduril⁹. In missiles and munitions, Leidos, CoAspire and Zone 5 hold the same framework agreements¹⁶. Castelion is the company named for the hypersonic deal¹⁶. Among venture-backed peers, the press names Shield AI, Saronic, Helsing and Hermeus, with Helsing at an $18 billion valuation in July 2026¹².

The incumbents matter most. Lockheed Martin has $79 billion of trailing revenue²⁰. Anduril’s reported 2026 revenue target of $4.3 billion is a small share of the market, and the large primes control the supply chains, the classified programmes and many of the relationships. They can also buy or partner with start-ups. Anduril names several of them as partners, for example General Dynamics Electric Boat and HII as the final assemblers of its submarine parts¹¹.

IVThe moat and how it could be disrupted

I rate the moat as contestable. The advantages are real. Anduril has a fast product cycle, its own factories, a software layer already embedded in Army systems, and the trust that comes from winning several large programmes at the same time. Switching costs in defense can be high once a system is fielded, trained on and certified. But none of these has yet been tested through a full cycle of price pressure, program loss or a downturn in the budget.

Several protections are weaker than they look. The $20 billion Army deal is a purchasing vehicle, and the Army says the deal does not replace competition on future programmes⁴. The Air Force picked two aircraft makers on purpose so that it keeps a second source¹². Mission autonomy software for the drone wingmen is a separate piece that a buyer can source on its own. In the Fury flight tests the aircraft switched between Anduril’s autonomy software and Shield AI’s⁵, which shows how swappable the brain can be. Other buyers can copy the approach, and the Pentagon is actively recruiting new entrants¹⁶.

The business is also hardware heavy and still loses money on the numbers I can see. A disruptor could be a competitor with cheaper machines, a prime contractor that learns software, or a buyer that insists on open standards so that Lattice is one choice among many. Regulation and politics cut both ways. Defense is protected from foreign competition by export rules and security clearances. Anduril is also tied to political ties and to decisions it does not control: co-founder Palmer Luckey was named to a Pentagon initiative called Project Meridian days before the Navy award, and the Pentagon’s technology chief said there was nothing to see in the potential conflict-of-interest questions⁸. I cannot judge how that plays out under a different administration.

VThe numbers we can see

Anduril does not publish audited accounts. Every revenue number below is either a figure the company gave in a letter that the press reported, or a figure that the press reported from anonymous sources. I have found no SEC filing from the company itself. A search of the SEC’s EDGAR company database for registrations by Anduril returned no matching companies²¹, and a search for Form D notices returned two special purpose vehicles named for Anduril that are series of an outside fund, not Anduril’s own filings, with no amounts shown²².

The reported figure for 2025 is $2.2 billion of revenue, double 2024, from CEO Brian Schimpf’s letter that accompanied the May 2026 round¹¹⁰. Sacra puts 2024 at about $1 billion⁷; that is an estimate. For 2026, Contrary Research and Sacra both cite The Information for a target of about $4.3 billion and an operating loss of about $1.2 billion⁶⁷. That is press-reported. Contrary calls it both a doubling and about 50 percent growth in the same article, and 4.3 over 2.2 is about 95 percent, so I use the doubling reading and flag the inconsistency⁶. Sacra says it does not expect adjusted EBITDA profit until 2030⁷. That is Sacra’s statement, not the company’s.

Backlog is a different matter. I saw conflicting numbers in search summaries, one over $10 billion and one over $900 million, and neither came from a source I opened. I therefore give no backlog figure. The chart below shows only the two revenue points that I can source, and the 2024 point is labelled an estimate.

Valuation history is clearer. The June 2025 round was $2.5 billion at $30.5 billion, led by Founders Fund¹. The May 2026 Series H was $5 billion at $61 billion¹. TechCrunch says pre-money and post-money were not stated, and that its own subtraction implies about $56 billion before the new money¹; I treat $61 billion as the post-money figure on that reading but I cannot confirm it. On 13 October 2026 an employee tender opened at $94.80 a share, which implies about $90 billion, capped near $1.3 billion and open until 9 November¹⁴. Reuters reported in July that the company was in talks for a round near $100 billion, with Anduril saying no decisions had been made². A summary of an October report from The Information says the company hopes to raise at about $120 billion in the first quarter of 2027¹⁵. I did not open The Information itself. These are talks, not completed rounds.

Valuation marks and what each one is

VIWho runs it and who owns it

Brian Schimpf is chief executive and Trae Stephens is executive chairman²³. Palmer Luckey is the founder, and Matt Grimm, a co-founder, is chief operating officer¹²²³. Chris Brose is president and chief strategy officer⁸.

Owners include venture and growth investors. PitchBook, as relayed by TechCrunch, lists Thrive Capital, Andreessen Horowitz, Founders Fund, ICONIQ, Flux Capital, Greycroft, Altimeter and 1789 Capital, and names US Vice President JD Vance as an investor². I could not find a cap table, the size of the founders’ stakes, or the rights attached to each class of stock. That matters. Late-stage private shares often carry preferences that put some investors ahead of others, so a headline valuation is not the same as what an ordinary shareholder would get.

The tender is a secondary sale, which means existing holders sell to buyers and the company raises no new money in it¹⁴. It tells us what some buyers are willing to pay for existing shares. It does not show that new cash is going in at that price. The size is small, about $1.3 billion against a valuation of about $90 billion, so a high price is easier to hold up than it would be for a larger sale.

VIIValuation: is the last price supported?

Equity value in US$ billions. Not a share price: a private company has none. Latest mark: secondary (employee tender, implied); last primary round $61B post-money in 2026-05 (reading unconfirmed).

What the last valuation assumes

At $90 billion, with my base 5x terminal multiple, 14% discount rate and 12% dilution, I need about $34 billion of revenue in 2030 for the price to be fair today. That is nearly eight times the 2026 target of $4.3 billion⁶, or about 67 percent growth a year for four years. At $61 billion I need about $22.5 billion in 2030. The arithmetic is in my model script. A different terminal multiple moves this sharply: if the market treats Anduril as a software company, the multiple could be much higher than 5x and the requirement far lower.

Bear, base and bull

• Bear. 2030 revenue $8B (about 17% a year from the $4.3B 2026 target), because the aircraft programme slips and the Army ceiling is little used. Terminal EV/sales 3.0x, the range of AeroVironment (3.3x) and Kratos (4.1x)¹⁹¹⁸. Discount rate 15%. Dilution 12% over the period. Net cash $3B (my estimate from the $5B May raise, partly spent).
• Base. 2030 revenue $14B (about 34% a year). Terminal EV/sales 5.0x: above Kratos (4.1x) because of software and growth, far below Palantir (57x)¹⁸¹⁷. Discount rate 14%, a high-risk equity rate (my estimate). Dilution 12%. Net cash $4B (my estimate).
• Bull. 2030 revenue $20B (about 47% a year), with both factories at rate and software revenue growing. Terminal EV/sales 7.0x, a blend of the defense manufacturers and a software premium. Discount rate 12%. Dilution 10%. Net cash $4B.

Comparables used

Where I come down

The latest valuation is above my base case. My base is about $40 billion, with a range of $15 billion to $82 billion. The $61 billion May round is about 1.5 times my base, and the $90 billion tender is more than twice it. To justify either price I would have to believe in revenue far above the 2026 target, a terminal multiple well above the defense manufacturers, or both. The market could still be right. The company doubled revenue in 2025, won a $20 billion ceiling contract, a Navy award and a production decision within a few months⁴¹¹¹², and the buyers in these rounds may be paying for a rare asset in a sector that governments want to expand. Prices in a hot private market can also run ahead of fundamentals, and a tender with a $1.3 billion cap is a small test. My model is simple and uses one revenue path with a few cases, so I do not lean on it heavily. The margin of safety figure, 25 percent below base, is about $30 billion.

What I cannot judge

I cannot judge gross margin, backlog, the split between hardware and software, the terms of preferred stock, the cap table, or how much cash the company holds. I could not open Anduril’s own website or the original Bloomberg and The Information pieces, so several figures come via secondary coverage. I also cannot judge the technical performance of the weapons, which is mostly classified. The comparables’ multiples come from a market-data site as of early November 2026 and were not checked against filings. Public data is too thin for a precise value, so treat the range as wide.

VIIIThe IPO

No IPO plan reported. No S-1 appears on SEC EDGAR for Anduril²¹. Co-founder Luckey has said an IPO is certain and necessary to compete for larger contracts, but has given no timeframe¹⁰¹⁴. Press coverage in May 2026 said investors widely expect a listing¹⁰. A secondary sale at about $90 billion¹⁴ and reports of a 2027 raise at a higher price¹⁵ point to continued private funding. I found no named banks, exchange or ticker. Sites that claim a ticker or date are unsupported. I read the status as no plan reported, not as ruled out.

What to watch

• A public or confidential S-1 filing, or credible reports of banks hired
• Whether the tender closes on 9 November 2026 at $94.80 and is fully subscribed¹⁴
• Whether a round in early 2027 prices near $100 to $120 billion²¹⁵
• Arsenal-1 delivery rate against its capacity of up to 150 aircraft a year, and which autonomy software the Air Force picks¹²
• How much of the $20 billion Army ceiling becomes actual orders⁴
• Audited or filed financials, if any appear

IXWhat could go wrong

I start with what most likely hurts an owner. The first risk is concentration on one customer. Nearly all revenue comes from governments, led by the US, and a ceiling contract can sit unused if budgets or priorities change. The second is factory execution. Arsenal-1 had a first aircraft in July 2026, but at the rollout the main hangar was still unfinished and a second building was a shell¹². Arsenal-2 is a $3.7 billion bet that first operates around 2030¹¹. A company that loses $1.2 billion on $4.3 billion of revenue while also building shipyards needs outside money for years⁶¹¹.

The third risk is price. At $61 billion the company is valued at about 14 times its 2026 revenue target, and at $90 billion about 21 times, using the reported $4.3 billion¹⁶¹⁴. By contrast the listed comparables are far lower. Lockheed Martin is at 2.1 times trailing sales²⁰, AeroVironment 3.3 times¹⁹ and Kratos 4.1 times¹⁸. Palantir is at 57 times¹⁷. A buyer at $90 billion is saying Anduril looks more like Palantir’s software economics than a defense manufacturer’s.

What has to go right at the same time? In my base case, revenue must grow about 34 percent a year from 2026 to 2030, the Air Force production programme must reach rate, the Army software and the munitions programmes must convert from ceilings into orders, the factories must deliver at the planned cost, and margins must turn positive without a funding squeeze. If I give each of four things, say, a 70 percent chance and treat them as independent, the combined chance is about 24 percent. They are not independent, since they share one budget and one management team, so that number is a rough guide to how many things need to line up, not a forecast.

The next best alternative for the same dollar is a listed defense name or a diversified fund. They give up growth and offer audited accounts, trading in public markets, and cash returns. A private holder generally has to wait for an IPO or a sale, and for most retail holders there is no easy way to buy at all. I cannot compare returns because the evidence does not allow it. A fair test for any holder is what the business could pay out once the building phase ends. With a reported 2026 operating loss and two large factory projects, I see no basis to estimate cash that an owner could take out before the end of the decade, and I put none in the model beyond the terminal value. Also note the political risk. Luckey’s role in the Pentagon’s Project Meridian drew conflict-of-interest questions⁸, and in July 2024 China’s foreign ministry imposed countermeasures on the company and three of its senior executives, citing US arms sales to Taiwan³².