Northrop Grumman

IThe business

Northrop Grumman builds military hardware that has no commercial twin: a stealth bomber, an intercontinental missile, missile-warning satellites, solid rocket motors, radars. In 2025 it had $41.95B of sales, 2% more than in 2024, about 95,000 employees, and four sectors.³ The US government bought 84% of the sales, 87% in 2024 and 86% in 2023; in the first nine months of 2026 it was 85%, with 14% international.³⁵ The head office is in Falls Church, Virginia. Its customer is called the Department of War in the filings, and I use that name.

It has not always looked like this. It started as Northrop in 1939 and became the company it is by buying: Grumman in 1994, Westinghouse’s defence electronics in 1996, Litton and Newport News Shipbuilding in 2001, TRW in 2002.³ Then it shed. The shipbuilding business was spun off as Huntington Ingalls in 2011, and the IT and mission support services business was sold to Veritas Capital in 2021 for proceeds of $3.4B and a gain of $1.98B.³⁹ In 2018 it bought Orbital ATK for $7.7B, which added solid rocket motors, satellites and launch vehicles.³⁹ The shape that remains is hardware: it is capital-heavy, it is hard to copy, and it is paid by one customer.

How it earns money

Most of the work is done under long contracts that the company accounts for as the costs are incurred. On a cost-type contract the customer pays the costs and a fee, so the risk of a surprise sits with the government. On a fixed-price contract the company is paid a set price, so a surprise comes out of its profit. In the first nine months of 2026, 52% of sales were fixed-price, up from 49% a year earlier, and 48% were cost-type.⁵ The mix is moving the wrong way for anyone who wants less risk, and it is moving by policy: the Administration has issued orders to promote fixed-price contracts where appropriate, and the company lists that among the risks.⁵ Every quarter the company revises its estimate of what each contract will cost, and the change is booked at once. In 2025 the favourable revisions were $1,696M and the unfavourable $1,487M, a net of $209M; in 2024 the net was $350M.³ Those are the quiet gains and losses. The loud ones are the two or three programmes where a single revision can run into the billions, and they are the subject of most of this page.

The B-21 Raider

The B-21 is the new long-range stealth bomber for the Air Force. Northrop Grumman won the contract in 2015. It has a development phase, which is largely cost-type, and five low-rate initial production lots, which are largely fixed-price, for a baseline of 21 aircraft, with the lots expected to be awarded through about the end of the decade.³ The price of those five lots was set before anyone knew what it would cost to build the aircraft at a steady pace. That is the whole story.

In the fourth quarter of 2023 the company booked a loss of $1.56B across the five lots. In the first quarter of 2025 it booked another $477M, which cut the sector’s operating margin for the year to 6.3%, and the shares fell 12.7% on the day the results came out.³¹ Together the two charges are about $2.0B, and the company says so.⁵ It reviewed the lots again in the third quarter of 2026 and made no significant change.⁵ An accounting loss is booked when the whole contract is expected to lose money, and the cash goes out afterwards, as the aircraft are built. The remaining loss accrual was $1.3B at the end of 2025 and $1.0B on 30 September 2026, so about $0.3B of the cash cost came out of the business in nine months, and about $1.0B still has to.³⁵ That money is not in any profit figure I am going to use in the future. I treat it as a debt.

The company and the Air Force have also set not-to-exceed prices for further lots, up to unit 40 in the 10-K and two additional lots in the latest 10-Q. The average of those prices is above the five-lot average, and they carry a clause that adjusts for some inflation. The terms are not fully negotiated.³⁵ In the first quarter of 2026 the two sides agreed to expand production capacity and raise the production rate. The company expects to spend about $2.5B over several years to do it and says, in return, it has the opportunity to earn improved returns on the production lots.⁵ The word is opportunity. Nothing in the filing promises a margin. It also sold the Air Force an aircraft it had planned to keep as a company test asset; the $371M of cash came in the third quarter and is classified as investing, not operating, which flatters the company’s adjusted free cash flow for 2026 by that amount, and I take it out.⁵

What does this mean for an owner? It means that the best-known programme in the company is currently a loss, that the loss is mostly behind it in the accounts but not in the cash, that the company is spending $2.5B to make more of an aircraft on which it has lost money, and that the payoff is a margin that nobody outside the Air Force and the company can see. It does not mean the programme is bad. A bomber nobody else can build, bought in numbers for decades, is the kind of asset I want a company to own. It means I cannot know the return, so I give it a range, and my bear case adds a further $2.0B of cash losses across the B-21 and Sentinel over 2027 to 2030.

Sentinel, the missile that breached

Sentinel is the modernisation of the country’s land-based intercontinental ballistic missile force, the ground-based leg of the nuclear triad.³ The Air Force awarded Northrop Grumman a $13.3B contract for the development phase in 2020. In January 2024 the Air Force told Congress that the programme was under a Nunn-McCurdy review, a legal check that is triggered when the estimated cost of a major programme grows past set limits; the growth came mostly from the production and deployment phases, and mainly from the new command and launch infrastructure.⁵ In July 2024 the Department certified the programme to continue and directed that it be restructured.⁵ I did not open the Department’s own statement of how large the growth was, and I will not quote a number I have not read.

The restructure is still under way. It includes a new acquisition strategy, a new baseline, and a repeat of the Milestone B approval, the formal step that lets a programme go forward.⁵ The company is working under a cost-type development contract, so the cost growth sits with the government for now, but the production and deployment phases have not been priced.⁵ That is where the risk sits. If the restructured production phase is priced as a fixed price with a thin margin, Sentinel becomes the next B-21. In the second and third quarters of 2026 the company and the Air Force authorised more of the plan, which added $9.4B to the programme’s backlog.⁵ Sentinel sales rose $356M in the first nine months of 2026, and a $76M favourable estimate change in 2025 helped the Defense Systems margin.⁵³ The honest reading is that Sentinel is a very large, very long contract that the country cannot cancel without losing its land-based deterrent, and that its price is still being decided.

Space, missile defence and the F-35

Space. Space Systems had $10.77B of sales in 2025, 8% less than in 2024, because work on a restricted space programme and on the Next Generation Interceptor wound down, which cost $738M, and the timing of materials delayed satellites for the Space Development Agency by $172M.³ The sector builds the Tracking and Transport layers of that agency’s satellite network, the Next-Gen OPIR missile-warning satellites, the Cygnus cargo spacecraft, the HALO module for the lunar Gateway, and the GEM 63 solid rocket boosters used on the Atlas V and Vulcan rockets.³ The GEM 63 has a bigger cousin, the GEM 63XL, and in the second quarter of 2026 the company took a $91M unfavourable estimate change on it, with sales on that programme $159M lower in the first nine months.⁵ The 2025 awards included $2.5B for GEM 63.³ Space backlog is $26.3B, almost unchanged this year, and the sector’s nine-month operating margin is 9.8%, down from 10.9%.⁵ This is the sector that most needs a better year.

Missile defence and weapons. Defense Systems has the Integrated Battle Command System for the Army and Poland, the Glide Phase Interceptor, the third stage of the Standard Missile-3, and the rocket motors for the Precision Strike Missile and the GMLRS.³ The 2025 awards included $1.8B for the Ground-based Midcourse Defense system.³ Its backlog grew 31% in nine months to $36.5B, largely Sentinel.⁵ The same sector took a $68M unfavourable change on the Stand-in Attack Weapon in the second quarter of 2026, an air-launched missile for the F-35.⁵ I did not find, in either filing, any statement about a role in a new national missile-defence programme, so I leave that out.

The F-35. The 10-K says Northrop Grumman does fuselage production for the F-35 and for the F/A-18, and Mission Systems makes the F-35 radar, the distributed aperture system and the communications, navigation and identification avionics.³ F-35 sales rose $385M in 2025 and another $229M in the first nine months of 2026, and the 2025 awards included $3.3B for F-35 work.³⁵ The company does not say what share of its sales is F-35, and I will not guess. One fact from outside matters. Lockheed Martin, which builds the whole aircraft, said on 22 October 2026 that it was still negotiating production lots 20 and higher and that its costs had begun to exceed the contractual authorisation and funding on the Lot 20 advance contract.⁷ A supplier is paid after the prime is. If the prime’s contract stalls, the money for the fuselage can too.

The state of play now

The third quarter of 2026 was a good quarter. Sales rose 7% to $11.12B, segment operating income rose 5% to $1.35B, and diluted earnings per share were $7.89, against $7.67 a year earlier.⁴ Net awards were $16.4B, 1.5 times sales, and backlog reached a record $109.6B, 15% more than at the end of 2025.⁴⁵ On 20 October the company raised its 2026 guidance: sales of $43.75B to $44.25B, segment operating income of $4.94B to $5.04B, earnings per share excluding the pension mark of $29.35 to $29.75, and adjusted free cash flow of $3.1B to $3.5B.⁴ I remove $371M from that last range, for the reason above. Aeronautics Systems grew 14% in the quarter and Space Systems 2%.⁴ The share price closed at $555.32 on 2 November, 8.7% above the close of 19 October, the day before the results.¹

IIThe moat

The test of a durable business is what it does when it is under stress. Northrop Grumman has had two kinds of stress in the last ten years, and they do not give the same answer.

The first is the market. Sales grew from $24.7B in 2016 to $42.0B in 2025 and the backlog from $53.5B in 2018 to $109.6B, through a pandemic, a budget fight in most years, and a change of president.⁹⁵ Nobody took the B-21 or Sentinel away. A customer that has one supplier for a strategic bomber, and one supplier for the missile that replaces the Minuteman, does not walk. Defence is also the place where the customer’s own stated intent is to build more: the 2027 budget request is for $1.45 trillion, 44% above the 2026 level, though Congress has not acted on it.⁵

The second is the contract. The company did not lose the B-21 customer. It lost $2.0B on the production price. A moat that protects the position but not the profit is a franchise with a landlord, and the landlord is the government. On a fixed-price production lot the government has set the price in advance and bears none of the cost growth. So I ask a different question: are the profit and the return on capital good enough to survive the occasional charge? Operating income was $4.51B in 2025 even after the B-21 charge, which I count at 10.8% of sales, or 11.3% before the $477M charge and the $231M gain on the training business.³ The return on capital, which I calculate as operating income after a 17% tax over equity plus debt less cash, was 13.4% in 2025, and 36% on the capital outside goodwill, which is $17.4B of the total.³ These are good numbers for a business with this much fixed capital, not extraordinary ones.

What does protect it

• Incumbency on programmes that run for decades. The B-2 has been upgraded and sustained for years; the B-21 is the next bomber; Sentinel is the next land-based missile; the company also builds the GEM boosters and medium-class motors for the Trident II missile.³ A rival would need a new competition, and the government has not shown a wish to run one.
• Restricted work. About 40% of Aeronautics, 30% of Mission Systems and 45% of Space Systems is performed through restricted programmes, whose content is not public. That is a secret and a barrier at once.³ It is also a limit on what I can tell you.
• A backlog with a funding trail. $50.1B of the $109.6B is funded, which means authorised and appropriated; 35% is expected as sales in 12 months and 55% in 24 months.⁵ Few industrial companies can see their next two years so clearly.
• Solid rocket motors. The Orbital ATK purchase gave it motors for the GMLRS, the Precision Strike Missile, the Standard Missile-3, the GEM boosters, the Space Launch System and Trident, and sales on tactical motor programmes rose $114M in the first nine months of 2026.³⁵ I believe few firms make motors at this scale, but I have no market-share data, so I call that a belief, not a number.

Where it is thin

The company is one of five or six primes that compete for large awards, and the 10-K lists Boeing, General Dynamics, L3Harris, Lockheed Martin and RTX, and now, in its words, new entrants and startups.³ Competition happens at the award, not after. Protests of awards are common, and the loss of a single programme can empty a business: the wind-down of a restricted space programme and the Next Generation Interceptor took $738M out of 2025 sales.³ The company also has little pricing power. On cost-type contracts the fee is set by the government’s rules, and on fixed-price ones the price was agreed in a competition. The sector margins I find, 6% to 15%, are set by contract type more than by brand.

The balance sheet adds a second weakness. Goodwill is $17.4B, 34% of total assets, almost all of it from the purchases of 2001 to 2018.³ If a sector were to lose a franchise, the write-down would be large. The company has $15.2B of debt, and it paid for the Orbital ATK purchase with borrowing in 2018.⁵⁹ I do not read these as a present danger, because net interest paid was $588M in 2025 against $4.5B of operating income, but the time to worry about them is when a franchise is lost, not before.

IIIThe competitive advantage

The question here is what Northrop Grumman can do that a rival cannot. The right comparison is Lockheed Martin, which has the F-35, the biggest defence franchise in the world, and a backlog of $292.0B, and then General Dynamics, RTX and L3Harris. Each reports every quarter.

Two things stand out. First, size: Lockheed Martin’s backlog is 2.7 times Northrop Grumman’s, and its third-quarter sales were 1.9 times. Second, price. At $555.32, the shares trade at 18.8 times the middle of the company’s own 2026 earnings guidance, against 21.1 times for Lockheed Martin and 26.3 times for RTX, though the definitions of earnings differ, and Northrop Grumman’s includes a pension benefit of about $3.90 a share that is not cash, which is the nine-month benefit of $499M taken at four thirds and taxed at 17% C (calculated).^4711135 That discount to its peers is the reason the page is worth writing. It could be a mistake in the market, or a price for the B-21 and Sentinel risk that the others do not carry in the same size.

So where is the advantage? I find three places, and none is certain.

Strategic positions that cannot be rebid. Lockheed Martin has the F-35 and a $52.5B interceptor award; General Dynamics has submarines and Gulfstream.⁷¹⁰ Northrop Grumman has the aircraft that carries the nuclear mission, the land-based missile, and a large share of the satellites that watch for launches.³ These are programmes where a government cannot easily switch supplier, and where it pays for the capability whatever the supplier’s margin.

A mix of businesses that does not follow one budget line. Mission Systems earns a 14.6% margin and is the second-biggest sector; the sensors, radars and electronic warfare it builds go into other companies’ aircraft and ships.³ That is a lower-risk, higher-margin business than the bomber, and it is nearly 30% of sector sales.

No shipyard. The 2011 spin-off meant that Northrop Grumman does not build ships. General Dynamics’ Marine Systems earned a 7.9% operating margin in the latest quarter and 7.5% in nine months on $14.0B of sales, and its sales rose 17.3%, against Northrop Grumman’s company margin of 10.5% for the same nine months.¹⁰⁵ The company still supplies submarines, with power and propulsion systems for Virginia and Columbia class boats and a $1.4B Virginia award in the third quarter, but it does not carry a shipyard’s capital and labour risk.³⁵

Where rivals can catch up

The most direct threat is the new entrant. The 10-K now names startups as competitors, and the Department’s stated aim is to buy speed and low cost for some systems.³ The strategic bomber and the ICBM are not that kind of market. The sensors, satellites and drones are. If the next ten years of spending go to cheap, mass-produced systems, Northrop Grumman’s cost-plus franchise will grow more slowly than the budget. The second threat is the customer: the Administration has issued orders on fixed-price contracting, underperformance, production investment and incentive pay, and the company lists that as a risk to its business.⁵

IVManagement

I start with how the people who run it are paid, because that tells me what they will do when two good things conflict.

Kathy Warden has been chief executive since January 2019 and chair since August 2019; she joined in 2008 and came up through the Mission Systems and Information Systems sectors.⁸ In 2025 she was paid $25.36M: a salary of $1.84M, stock awards of $16.75M, an annual bonus of $4.22M, and $2.55M of pension change and other pay.⁸ The median employee earned $112,998, a ratio of 224 to 1.⁸ John Greene became chief financial officer in 2026; he was the chief financial officer of Discover Financial Services from 2019 to 2025, and he follows Kenneth Crews, the previous finance chief.³⁸ A new finance chief who comes from a bank is a fresh pair of eyes on programme estimates, which is where the company’s two big surprises came from. I cannot judge whether that helps.

What the pay rewards

The annual bonus for 2025 was weighted 35% to adjusted cash flow from operations, 35% to growth in adjusted segment operating income, 20% to adjusted operating margin and 10% to non-financial goals. Warden’s target is 200% of salary, and the company’s performance factor for 2025 was 114%, so she was paid 114% of target.⁸ The long-term awards pay in shares on three equally weighted measures over three years: adjusted cumulative free cash flow, return on invested capital, and total return against peers; they can pay from 0% to 200%.⁸ I like the cash flow and return measures. They are close to what an owner wants.

Two things in the fine print matter. The committee excluded several items from the bonus numbers, including federal and foreign tax payments and CAS pension recoveries, with discretion to adjust for conditions it calls unanticipated.⁸ And the return-on-capital measure for the long-term awards is adjusted for the fourth-quarter 2023 charge on the B-21 production lots.⁸ So the largest single failure of the last five years was taken out of the number that pays the managers. That is a normal practice, and the committee says it removes events that were not foreseen. But when the pricing decision that led to the charge was management’s, an owner needs to know that the pay for return on capital did not feel it. Nothing in the pay measures penalises a fixed-price bid that wins. The measures reward backlog only indirectly.

How much of their own money is in

The proxy lists 196,681 shares for Warden, worth about $109M at $555.32, which is 59 times her salary, against a guideline of 7 times.⁸¹ The other four named executives hold between 2,353 and 11,205 shares each. All directors and executive officers together hold 303,538 shares and share equivalents, about 0.2% of the company.⁸ The ownership by the chief executive is meaningful; by the rest it is modest. Shares received must be held for three years for half of the net after tax.⁸ The four holders of more than 5% are all index and institutional managers: State Street 9.8%, Vanguard 9.3%, Capital International Investors 8.7% and BlackRock 7.5%, as reported in filings dated between January 2024 and August 2025 and listed in the proxy.⁸ None of them controls the company.

What they did with the capital

Over ten years the company paid $9.6B of dividends and bought back $15.3B of stock, about 43 million shares at an average of $355, against a price today of $555.32.⁹ That was good timing on average: the purchases were mostly made below today’s price, and the share count fell from 180.5M to 143.8M, or 2.5% a year.⁹ The largest purchases were in 2021, $3.7B, funded by the sale of IT services at a gain of $1.98B.⁹ In the first nine months of 2026 the company spent $68M on buybacks, against $1.17B a year earlier, and nothing in the third quarter.⁵ The filings do not say why, and I do not guess. Cash went instead to repaying $527M of notes and to the dividend, while the company plans about $2.5B of B-21 capacity spending.⁵

The acquisitions record is shorter. Orbital ATK in 2018 cost $7.7B and brought solid rocket motors, satellites and launch vehicles into what are now Space Systems and Defense Systems; Space Systems earned 11.0% on $10.8B of sales in 2025, and I cannot separate the part that came from Orbital ATK.⁹³ The 10-K’s list of acquisitions ends there. That is a quieter record than most companies of this size, and it leaves the larger question unanswered: whether the next use of $2.5B, the B-21 expansion, earns more than it costs.

The legal record has one open item. Since 2019 the Defense Contract Management Agency has questioned the interest-rate assumption the company used for its CAS pension costs for 2013 to 2019, and the Justice Department issued a criminal subpoena in December 2022 and a civil demand in February 2023. On 20 May 2026 the Criminal Division closed its inquiry. The civil demand and the agency review remain, and the company says the outcome could be material and gives no range.⁵

VFinancials

Ten years of numbers, as filed. They are not clean, and I will say how.

Sales grew 6.1% a year from 2016 to 2025, with the biggest step from Orbital ATK in 2018; 2021 fell 3% because the IT services business left.⁹ Operating margin was 13.3% in 2016, 12.6% in 2018, and 10.8% in 2025, and in between it was 15.8% in 2021 because of the $1.98B gain and 6.5% in 2023 because of the B-21 charge. Part of the step down from 2016 is the amortisation of the Orbital ATK intangibles and part is the B-21; the company does not publish the pieces, so I cannot split them. Net earnings and earnings per share jump around for the reasons set out in the chart: pension marks, gains and charges.

Backlog grew from $53.5B in 2018 to $109.6B in September 2026, doubling in eight years, with the latest nine months up $13.9B, mostly Defense Systems, where Sentinel and missile awards sit.⁹⁵ Backlog is not profit, and it is not all fixed: unexercised options are not in it, and the funded part is $50.1B, which is 46%.⁵

Earnings are not cash

Reported earnings are a poor guide here. Two pension items sit in them: a mark-to-market gain or loss booked in the fourth quarter, a gain of $527M in 2025, and an annual non-operating pension credit of $541M, which comes from the surplus in the plan and is not cash.³ Remove both and 2025 earnings were $3.34B, not $4.18B, and owner cash was $3.19B.³ That is a smaller gap than I expected, and it needs one caution: owner cash was helped by a lower cash tax bill after the 2025 tax law allowed research costs to be deducted at once, and it was hurt by working capital. In the first nine months of 2026, trade working capital took $2.15B of cash, and operating cash flow was $1.37B, because the fourth quarter is when the cash arrives.⁵ Operating cash flow in 2025 was $0.86B after nine months and $4.76B for the year.⁵³ So I do not judge the cash flow from three quarters.

Capital spending and what it takes to stand still

Capital spending was $1.45B in 2025 against depreciation and amortisation of $1.47B, so the company is spending about what the business wears out.⁹ It was $1.77B in 2023 and 2024. The $2.5B of B-21 capacity spending is on top of the base and is spread over several years, though I do not know in which years the cash goes.⁵ I take capital spending at 3.4% of sales in 2027 and 2028 in my base case, $1.6B and $1.7B, which leaves room for part of that spend. Stock pay was $119M. That is a small cost next to the $3.19B, and I still deduct it.

The pension needs a plain statement. The plans hold $1.7B more than they owe on the company’s accounting, with a required cash contribution of $97M in 2026. That is good news for cash, and it means the company is not going to write large cheques. But the surplus is why the earnings carry a $541M credit, and the credit will not last forever. The pension cost charged to contracts, which the government pays, is the other side: CAS pension expense was $475M in 2025, higher by 70%.³ The FAS/CAS adjustment, the gap between what is charged and what is booked, fell to $29M in nine months of 2026 from $193M a year earlier, which cost the company margin.⁴

Debt is $15.2B, 3.4 times 2025 operating income, with a ladder that has $755M due in 2027 and $2.0B in 2028, at rates of 3.2% and 3.25% that will be refinanced at higher ones.³⁵ Net interest paid was $588M in 2025.³ I do not see a funding risk. I do see that the cheapest debt is the part that rolls first.

The shares closed at $555.32. The highest close in ten years was $768.02 on 2 March 2026, and the price is 28% below it. On 7 October 2026 it closed at $473.46, the 52-week low.¹ Since the close before the third-quarter results the shares have gained 8.7%. In ten years the price has gone from $228.67 to $555.32, a rise of 9.3% a year before dividends.¹

VIValuation

First, how this goes badly wrong

Before any numbers I ask what would hurt an owner most. I count three ways.

A second B-21. The company has now lost $2.0B on five fixed-price lots, and the next lots, the ones that are meant to repay it, are priced under not-to-exceed terms that are not fully negotiated.⁵ Sentinel is the other. Its development is cost-type, but the production and deployment phases are not yet priced, and the policy of the day is to prefer fixed-price contracts.⁵ If the restructured production price is set thin and costs run over, one programme could cost another billion or more, and a charge is a cash loss. My bear adds $2.0B of such losses across 2027 to 2030. The 2023 and 2025 charges were $1.56B and $0.48B. They came without notice, on the day of a quarterly report. A $1.5B charge is only about $10 a share. What hurts more is what it says about the margin on every lot after it.

The money stops, or stops on time. 84% of sales are to one customer, and the customer’s budget is passed year by year. On 2 November 2026 the government is running on a continuing resolution that lasts to 11 December, which forbids new multi-year contracts without approval from Congress, and the 2027 request of $1.45 trillion is not law.⁵ The company’s guidance assumes no prolonged shutdown.⁴ A late budget delays awards and payments more than it cancels programmes, and that is why the damage tends to show up in working capital and in awards before it shows up in sales. I cannot forecast Congress.

I pay a fair price for a good record and the record is flattered. The shares trade at 18.8 times guided earnings, but those earnings include a non-cash pension credit of about $3.90 a share, and 2025 and 2026 cash is helped by a one-off tax change. Take away the pension credit and the multiple is 22 times; take away the sale of the B-21 test aircraft and the free cash flow yield is about 3.7%. A defence contractor that pays out most of what it earns, and has a growth rate near 5%, is not a bargain at 22 times, though it is a better business than most.

What an owner can really take out

In 2025 owner cash was $3.19B, or $22.4 a share, 4.0% of the price C (calculated).³ In 2026 the company’s guide for adjusted free cash flow is $3.1B to $3.5B; it includes $371M for the B-21 test aircraft, which is a one-off, so the midpoint comes down to $2.93B and then to about $2.83B after stock pay, which is $19.9 a share, or 3.6% of the price E (estimate).⁴⁵ In my base case, owner cash before interest and dividends is $3,994M in 2028 and $4,678M in 2030. Take off $0.53B of after-tax interest and an owner has about $4.1B in 2030, or $29 a share E (estimate). The dividend is $9.88 a share. Reported earnings per share of $29.55 at the middle of guidance are above any of these; the gap is the pension credit, a low tax rate and the working capital that the company says it will take back in the fourth quarter.

What I can’t judge

I can’t say what the B-21 will cost to build. The Air Force and the company know it, and neither publishes it. I can’t say whether the improved returns the company expects from the production deal are 2 points of margin or 8 points on those lots. I can’t say how large the Sentinel production price will be, or on what terms, or what the Department’s own estimate of the restructured programme is, because I did not read the Department’s statement. I can’t see inside the restricted programmes, which are 40% of Aeronautics, 30% of Mission Systems and 45% of Space Systems.³ I can’t say what a fourth-quarter pension remeasurement will do; it is a swing of hundreds of millions in either direction and it does not move cash. And I can’t say what Congress will do in December. So I show a ladder of margins and not one number.

How the cases are built

I build each case year by year in US dollars. Sales for 2026 are $44.0B, the middle of guidance, and grow from there. The operating margin is total operating income over sales, with pension on the accounting basis and without the non-operating pension credit. The 2026 segment guidance is 11.3%, and the company-level margin in nine months is 10.5%; I start 2026 at 10.6%. Cash tax is 16.5% of operating income in the base, below the 17.5% of 2025 because of research credits. Depreciation is 3.3% of sales, capital spending 3.4% falling to 3.2%, and working capital takes 10 cents of every extra dollar of sales. Stock pay of $125M a year growing 4% is deducted. Debt of $15.2B, pension liabilities of $1.08B and the remaining B-21 loss accrual of $1.0B are claims ahead of the shares, less cash of $3.74B, a net $13.5B.⁵ Cash from the fourth quarter of 2026 counts in the model; it is the full-year estimate above, less nine months of actual cash, and it is large because the fourth quarter is when the cash comes in.⁵⁴ I discount at 8.5%: the 10-year Treasury at 4.94% on 30 October plus 3.6 points for a customer concentration of 84%, programmes with unknown costs, and a budget I cannot forecast.⁶ The premium is smaller than I would ask of a company with one product line, because the customer is the best credit there is, and larger than I would ask if its programmes had no unknown costs. I stop in 2036 and then grow the last year’s cash at 2.5% in the base case.

Three comparisons help. My base 2026 operating income is $4,664M, against 2026 segment guidance of $4.94B to $5.04B and nine-month operating income of $3.35B.⁴ My base then grows 4.7% a year to $7.4B in 2036. That is modest next to a backlog that has doubled in eight years, because I let growth fall to 3% by 2034 and hold the margin at 11.0% from 2030. My bull reaches $7.0B in 2030 on a margin of 12.1%, which needs the B-21 production deal to pay off and Sentinel to be priced fairly. My bear has sales growing 2% to 3% a year and margin falling to 8.8% in 2028 from the extra losses before it recovers to 10.2%.

In plain words, the price needs one of three things beyond my base: an operating margin of about 12.6% held for ten years, which is above the 11.3% of 2025 before the charge and the gain, and above every year since 2018 apart from the 2021 gain year; or about 10% more sales by 2030 than I assume; or a discount rate of 7.7%, which is the return on a safe but not risk-free asset. None of these is silly in isolation. Together they say the market is paying for the B-21 to be the success it has not yet been, and for budgets that keep growing E (estimate).

The margin of safety, and the price against the range

My margin-of-safety price is $328. It is 70% of my base value. At that price an owner would have a 30% cushion against being wrong about the B-21, Sentinel and the tax, and it would be 11% above my bear. It is about 14.6 times 2025 owner cash and 11.1 times the middle of 2026 guided earnings. The shares have not closed at or below $328 since 1 April 2021; the lowest close of the last year was $473.46, and the close after the B-21 charge in April 2025 was $464.08.¹ At $555.32 there is no margin of safety. The price is 19% above my base, 10% under my bull and 1.9 times my bear. My weighted value, with 25% on the bear, 50% on the base and 25% on the bull, is $463.

The price reflects a real debate. The case for the market being right says that Northrop Grumman is the sole supplier on three of the largest strategic programmes in the country, that its backlog has grown 15% in nine months to a record, that the government is asking for $1.45 trillion, that the loss on the B-21 is booked and the improvement is still to come, and that the company is cheaper than its peers at 18.8 times. If I give all of that its weight, my bull case is where the price goes. If the shares reach that and the safe rate falls to 7.5%, my base value rises to $579, which is above the price. The case against says that the operating margin has been below its 2016 level in every year since, except 2021, when a gain lifted it, that the B-21 and Sentinel have not finished costing money, that the cash flow is flattered by tax, and that I could be wrong by a billion in either direction. I give the second side a little more weight, and I would not be surprised if the first side were right for a while.

The next best use of the same dollar

The simplest alternative is the 10-year Treasury, at 4.94% on 30 October, with no budget risk and no charge risk.⁶ Against that, my base return to October 2036 is 6.9% a year including the dividend, 2.0 points above the bond and 1.6 below what I ask. The nearest listed alternative is Lockheed Martin, at 21.1 times its guided earnings with a free cash flow outlook of $7.8B to $8.0B on about 231.5M shares, a yield of 5.3% at $643.15, against 3.6% for Northrop Grumman on my arithmetic.⁷¹³ Lockheed Martin has its own open item in the F-35 contract. On this site I have not found a defence page to compare with, and I leave it there. Northrop Grumman is a good business. The price is a separate question. A good business at a price that needs a better margin than it has ever kept is a fair purchase at best, and I give neither a grade.

The bull case needs many things at once, so I multiply them out. Sales growth above 6% a year for four years, 30% likely by my guess; the B-21 production deal paying a margin that lifts the company above 12%, 25%; no new charge on the B-21 or Sentinel, 55%; a defence budget that grows in real terms, 60%; and no cash-tax surprise, 70%. Together that is about 2% if they are independent. They are not, since a bigger budget helps three of the five, so I call it 8% to 12%, not 25% E (estimate). My 25% weight on the bull is generous on that arithmetic.

Where this could be in 5 and 10 years

For the base to be right at five years, sales have to be near $57B in 2031 and operating margin near 11.0%, and the value is then $543 a share, which is below today’s price, because five years of cash have gone out and the rest has grown. At ten years the base reaches $630. In the bear the value is $394; in the bull $883. The inputs that drive it are margin, the extra programme losses, the budget and the 8.5% rate. These are estimates of value, not forecasts of the share price. The market price can sit above or below value for years, and in the last year it has ranged from $473.46 to $768.02.¹

VIIThe case for the price

What most people miss

Four things cut in the company’s favour and get little attention. First, the B-21 loss is largely booked. The company says it has recognised about $2.0B of loss on the five production lots, and the later lots carry an average price above the earlier ones with a clause for inflation.⁵ If that holds, the programme’s worst accounting is behind it, though the cash is not. Second, the pension is $1.7B overfunded, with $97M of required contributions in 2026.³ Third, the Justice Department’s criminal division closed its inquiry into the pension cost assumptions on 20 May 2026.⁵ Fourth, the order book is not only large but cost-heavy in a safe way: 48% of sales are cost-type, where the customer carries the overrun.⁵

The bull case

In plain words: the company holds the country’s new bomber and next missile, the money behind them is rising, the backlog is the largest it has been, and the first big loss is behind it. On 20 October 2026 Kathy Warden said she was confident that the record backlog, the portfolio and the investments would keep driving sales and earnings growth.⁴ The company raised its guidance for segment operating income and earnings in the same release.⁴ It says its investments provide the capacity for durable growth.⁴ The Administration’s 2027 request is for $1.45 trillion of defence money, 44% above the 2026 level, and the 2025 tax law supplied about $150B of mandatory defence funding available to 2029.⁵ The restructured Sentinel adds $9.4B of backlog, and the shares are cheaper than Lockheed Martin’s at 18.8 times guidance. That is the argument.

I looked for named analysts and investors who have published a view on this stock that I could check, and I did not find one I can cite. The four holders of more than 5% report holdings and no views.⁸ So the bull case on this page rests on what management has said and filed, and on the facts.

Against reality

If the B-21 production lots earn their improved returns, Sentinel is priced fairly, and sales grow at 6% to 7% for four years, the company reaches my bull case of $620 a share and the price of $555.32 is possible. The price needs less than my bull and more than my base: an operating margin about 1.6 points above my base, or about 10% more sales by 2030, or a discount rate of 7.7%.

Today the company is partway. The nine-month operating margin is 10.5%, below the 12.6% the price needs in 2030, and segment margin is 11.2%.⁴ Sales are growing 5%, below my 6% for 2027.⁵ Backlog is at a record and book-to-bill was 1.5, which supports the growth.⁴ The B-21 improvement has not appeared in the numbers; the Aeronautics margin of 10.1% in nine months comes from "strong performance on certain mature production and sustainment programs", and the company guides the sector to a high 9% for the year.⁵⁴ Sentinel is not priced. The cash flow is helped by a tax change and a one-off sale. So the facts support the base more than the bull, and they have not yet separated the bull from the price.

The odds multiplied out in Valuation are about 8% to 12% for the bull as a whole. The price asks for slightly less than that.

What a 10x would take. Ten times $555.32 is $5,553, a market value of about $791B. At 20 times earnings that needs about $39.5B of net earnings, 9.5 times the $4.18B of 2025 C (calculated). At a net margin of 10%, close to the 2025 level, that is sales of about $395B, 9.4 times today’s and equal to 46% of the $859B the government has appropriated for all of defence in fiscal 2026.⁵ In ten years it would take 26% a year. For comparison, the shares have risen 9.3% a year over the last ten years, from $228.67 to $555.32.¹ Tenfold from here is close to impossible, and I would not build anything on it.

Where I come down

What moved me was the franchise: three programmes the country cannot easily buy elsewhere, a record backlog and an overfunded pension, a record that shows a company that returns cash. What did not move me was the margin, which is lower than it was ten years ago, and the fact that the best-known programme is currently a loss that I cannot size from outside. What would raise my values is a Sentinel production price with a margin near the company average, an Aeronautics margin above 10% sustained, and cash flow without the tax help. Each half-point on the company margin from 2029 is worth about $24 a share.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest, and a test of each.

• Fixed-price contracting by policy. The Administration has issued orders to promote fixed-price contracts where appropriate, and others on underperformance, production investment and the incentive pay of defence contractors. The company says that carrying out these could hurt it.⁵ I cannot size it. The share of sales on fixed-price contracts is 52%, from 49%.⁵
• The F-35 contract. Lockheed Martin said on 22 October 2026 that it is still negotiating lots 20 and higher and that its costs were exceeding the authorisation on Lot 20.⁷ A supplier of fuselage is paid after the prime.
• The pension dispute. The civil investigative demand and the Defense Contract Management Agency review of the interest rate used for CAS pension costs are open. The company says an outcome could be material and gives no range.⁵ The criminal inquiry is closed.
• Programme write-offs on smaller programmes. In the second quarter of 2026 it took $91M on the GEM 63XL and $68M on the Stand-in Attack Weapon.⁵ Both are small. They show the pattern.
• Supply chain and labour. The filings list rare-earth and metals constraints, clearance-holding staff and 15 collective agreements covering about 4,300 employees, two of which come up in 2026.⁵³
• Cyber and new entrants. The 10-Q adds the risk of artificial-intelligence tools that find software weaknesses; the 10-K names startups as competitors.⁵³
• Goodwill. $17.4B, 34% of assets. A lost franchise would force a write-down that does not touch cash but would move the book.³

What a bad year looks like in the model

The surprise is how little a one-off costs. A $1.5B loss takes $9 off the base, because it is a single cash cost of about $10 a share. A delay of a year in sales, with the growth made up the next year, takes $1. What matters more is anything permanent: a budget that stops growing in real terms takes $74, and a thin Sentinel price takes $22. The risk is not a bad quarter. It is a decade of programmes that earn a little less than they should.

Nothing on this page is a forecast of the share price. It is an estimate of what the business is worth, and a record of why I think so.