Northrop Grumman NYSE: NOC

Northrop Grumman had fallen roughly 12% after earnings into a period of the largest announced defence-spending increases since 2009. Was the market pricing a contractor whose revenue is 85–90% locked into long-term government programmes for the demand it was about to receive?

Why the opportunity existed

The stock had fallen about 12% after earnings on worries about Sentinel and B-21 programme costs, inside a wider valuation reset that also hit Lockheed and RTX. Our argument was that the Street was underrating Northrop's long-term earnings power: consensus indicated flat revenue and very modest growth, and we thought growth would accelerate as the major long-term contracts moved into production and scaled up.

Forecast years are shown in a lighter weight. Present value of the explicit forecast period: 30,857.

Terminal value and the target

Shading is the implied return against the $473.20 share price at the time: green above, red below. The base case is outlined.

$ millions, LTM at 25 April 2025. Chosen multiple: EV/Revenue 2.7x, giving an implied $587.12 (+24.1%). EV/Revenue was selected as the primary metric because Boeing's negative operating income made EV/EBITDA and EV/FCF unusable across the set.

Entry price, April 2025

$473.20

Price target

$697.72

52-week high

$774.00 — target exceeded

Price, 18 March 2026

~$735, +29% year-to-date

Price, 14 September 2026

$527.39 — +11.5% against entry

Backlog, FY2024 (in the pitch)

$91.5B

Backlog, FY2025 close

$95.7B

Backlog, 2026

$104.7B

The demand thesis held and kept holding. What moved the stock down was the other side of the ledger. On 21 July 2026 Northrop raised 2026 sales guidance to $43.75–44.25B and MTM-adjusted EPS to $28.60–29.10, and the stock fell anyway: segment operating margin came in at 10.6% against 11.8%, segment operating income fell 5%, and the EPS beat rested on a 6.3% tax rate versus 17.7% rather than on operations, alongside a $68M Defense Systems adjustment and a $91M Space Systems charge on a rocket-motor redesign.

Forecast against outcome

2026 revenue, forecast in this model

$44,215M

2026 revenue, company guidance midpoint

$44,000M

Forecast error

under 0.5%, nine years out

EBIT margin assumed, 2025 → 2034

10.9% → 11.4%, drifting up

Segment operating margin, actual 2026

10.6%, down from 11.8%

As of 14 September 2026. Source: Public market data and Northrop Grumman disclosures.

Every figure below is checkable against the deck and against public filings. The arithmetic reconciles; the judgement calls are where the work is.

1. Error

   The L3Harris revenue figure is wrong, and it produces the comps conclusion

   The table carries L3Harris FY2024 revenue at $11,107.2M. The reported figure is $21.3B. Every other company in the set ties to its actual FY2024 revenue. The error makes L3Harris the highest EV/Revenue in the table at 6.7x — more than double the next name — which is what pulls the mean to 2.7x. The mean is the multiple that was selected. The median is 2.2x, which is exactly where Northrop itself traded. Corrected, the mean falls to roughly 2.0–2.1x. On a corrected comp set these comps imply Northrop was fairly valued, not 24.1% undervalued.

   Source: L3Harris Technologies full-year 2024 results, January 2025

2. Judgement call

   The exit multiple sits above every company in the comp set

   17.24x. The highest LTM EV/EBITDA in the same deck's comp table is Airbus at 16.0x; Northrop itself traded at 13.7x. The perpetuity method's implied 18.04x does not independently corroborate it, since both rest on the same 6.55% WACC.

3. Judgement call

   Terminal value carries 73% of the valuation

   $84,520M of the $115,377M enterprise value under the perpetuity method. WACC less terminal growth is 3.55%, so small changes in the discount rate move the output a long way. The deck's own sensitivity table shows it: at an 8% WACC and 3% growth the model implies $487.65, against a $473.20 share price. A 145 basis point move in WACC turns the call into a hold.

4. Error

   The market risk premium is mislabelled

   Cost of equity of 7.04% equals 4.31% plus 0.63 times 4.33%. The slide labels 8.64% as the market risk premium, but 8.64% is the expected market return; the premium actually used is the 4.33% difference. That is at the low end of conventional assumptions and it feeds a WACC only 224 basis points above the risk-free rate.

5. Error

   The EBIT series does not match the income statement in the same deck

   FY2022 appears as $6,338M on the income statement page and $3,601M in the discounted cash flow. Every year differs. If this is deliberate normalisation for pension accounting and one-off items it should have been stated and reconciled on the page; as published, the two tables disagree.

6. Judgement call

   The working capital forecast has no stated driver

   Change in net working capital moves from −$53M in 2027 to +$1,410M in 2028 and +$1,266M in 2029, then back to +$102M. Those swings drive free cash flow by −27% and +39% in the respective years, with no assumption behind them in the deck.

The demand thesis was right and stayed right. Backlog went from $91.5B to $104.7B, and the stock passed my $697.72 target inside a year, reaching $774.

Three things I would change. First, I set a price target with no time horizon and no sell discipline, so being right and knowing what to do about it were never connected.

Second, I forecast revenue off the defence budget and let margin drift upward, from 10.9% to 11.4%. My own income statement page showed operating margin at 10.6%, 6.5% and 9.8% across three years and EBITDA swinging from $8.7B to $3.9B to $6.8B. I put that volatility in the bull case rather than the risk case, and I sensitized WACC and terminal growth but never margin. When the stock fell in July 2026 it fell on margin and execution while guidance was being raised — the one line I had not stress-tested.

Third, I chose the mean of a comp set over the median without checking the outlier that produced it. That is a five-minute check I did not do.