Saab Q3 2025: When the Butterfly Thesis Became Testable
Inside the quarter where three exponential curves—geopolitics, software, and market structure—finally showed their economic signature.
TLDR
Q3 2025 is when Saab’s “butterfly thesis” became testable, not theoretical.
The Arexis AI contract finally exposes the software layer underlying Saab’s entire architecture.
The next 14 months decide whether Saab has wings—or is just a caterpillar with good marketing.
The Convergence No One Knows How to Value
Saab is experiencing a once-in-a-generation strategic inflection where three exponential curves intersect simultaneously. The first is geopolitical regime change, the shift from three decades of peace dividend back to permanent rearmament. The second is a technology paradigm shift from hardware-centric to software-defined warfare. The third is market structure transformation from project-based contract execution to platform economics with network effects.
This convergence is so rare that markets literally don’t have a framework to value it. The debate about whether Saab deserves to trade at 40 times earnings or 23 times EV/EBITDA misses the point entirely. You’re arguing about how much a caterpillar weighs while a butterfly is emerging from the chrysalis. The relevant question isn’t whether the current multiple is justified by traditional defense contractor metrics. It’s whether Saab is becoming something fundamentally different, and if so, what that’s worth.
I’ve been writing about this thesis for two quarters. Q3 2025 was the quarter it stopped being theory and became testable.
The EUR 258 Million Proof Point
From Bloomberg, the headline summary of Saab’s third quarter results:
“Saab AB raised its full-year sales growth guidance and reported third-quarter profit that beat estimates as the Swedish defense company benefits from a European rearmament drive following Russia’s invasion of Ukraine. The company now expects organic sales growth of 20-24% for 2025, up from a previous range of 16- 20%.”
21% YTD organic growth is impressive, and the guidance raise marks three consecutive years of 20-plus% expansion. But Saab has been beating and raising for long enough that this pattern is established. The quarter itself showed the expected breadth: Aeronautics grew 34% on Gripen deliveries, Dynamics delivered 19% EBIT margins, and the order backlog reached SEK 202 billion representing nearly three times trailing revenue. Revenue of SEK 15.9 billion and EBIT of SEK 1.37 billion for an 8.7% margin in what’s seasonally a softer quarter demonstrates the momentum.
What matters isn’t the Q3 numbers. It’s what Saab announced three weeks later when most investors had moved on to the next earnings cycle.
On November 14th , Saab won a EUR 549 million contract to provide its Arexis electronic warfare system for German Eurofighter jets. The deal splits into two tranches. The first is EUR 291 million for baseline integration work. The second is EUR 258 million for an AI-enabled version developed in partnership with Helsing, a European defense AI company. That second number is the most important figure Saab has disclosed in years.
Think about the strategic implications. Saab convinced Germany to install Swedish software and artificial intelligence into mission-critical systems of a fighter aircraft built by Airbus. This isn’t peripheral equipment. This is Saab becoming the software layer between the pilot and the aircraft’s electronic warfare capabilities. And the economics are revelatory: of the total EUR 549 million contract value, 47 % is explicitly software and algorithms.
This is the Rosetta Stone for decoding what Saab has been building. For two years, management has emphasized software-defined architectures, AI-powered systems, and continuous capability upgrades. CEO Micael Johansson talks frequently about deploying autonomous AI agents on Gripen fighters “in a matter of weeks” and claims no other fighter aircraft could match that pace of software deployment. But until Arexis, we couldn’t quantify what any of this meant economically.
Now we can. If 47 % of a major electronic warfare contract is software, apply that lens to Saab’s other revenue streams. The Gripen E is fundamentally a software-defined platform where the core advantage is rapid capability upgrades through code deployments rather than hardware modifications. The GlobalEye airborne early warning system sells for SEK 3 to 5 billion per aircraft, but the airframe is a modified Bombardier business jet Saab buys commercially. The value concentrates in sensor fusion algorithms, command and control software, and data processing capabilities. The counter-drone systems bundle Giraffe radars with targeting algorithms, threat classification AI, and swarm coordination software where the radar is increasingly the commodity and the software is the differentiator.
If I conservatively assume software represents 40 % of electronic warfare and sensor system revenue, 30 % of Gripen platform value, and 20 % of ground systems, while accounting for the more hardware-intensive portions of the business, I arrive at an estimate that software and systems integration is already 10 to 12 % of Saab’s total revenue. If those revenues carry 25 to 30 % EBIT margins versus the 9.3 % group average for the first nine months, software is contributing 25 to 40 % of total EBIT despite being only 10 to 12 % of revenue.
When analysts asked Johansson multiple times during the Q3 call to quantify software as a %age of sales, his answer was consistent: “I can’t do that today.” Not “we don’t break it out separately.” Not “it’s not material.” Just “I can’t do that today.” That phrasing matters. It suggests capability rather than materiality. He could quantify it. He’s choosing not to.
This is strategic concealment and I think it’s correct. Every quarter Johansson waits to disclose software revenue, Saab’s competitive moat deepens. Each new customer that integrates Saab’s architecture makes the next customer more likely to choose the same for interoperability. Each AI model improves with more training data from deployed systems. Each software deployment builds institutional knowledge competitors can’t replicate quickly. When Saab finally discloses these metrics at the February 2026 Capital Markets Day, it will be from a position of strength with established relationships and proven economics.
The post-earnings order flow reinforces the architecture. Colombia contracted for 17 Gripen E and F aircraft worth EUR 3.1 billion with deliveries through 2032, providing base-load utilization that makes Ukrainian capacity expansion economically viable if that financing materializes. The U.S. Army ordered Giraffe 1X radars for counter-drone missions, validating the sensor-agnostic network architecture. These aren’t random export wins. They’re three layers of the same system: platforms as vessels for software, sensors as network nodes, and decision-making layers that coordinate across both.
The Question That Determines the Valuation
The three curves converging is directionally correct. Geopolitical demand is structural not cyclical, as evidenced by SEK 120 billion of the backlog scheduled for delivery in 2026 and 2027. The technology shift is real, validated by Arexis economics and NATO selecting Saab to lead the Allied Underwater Battlespace Mission Network for interoperability standards. The market structure transformation has proof points in cross-platform adoption and systems integrator roles.
But one question determines whether Saab deserves to trade at 23 times forward EV/EBITDA as a defense contractor or 35 times as a software-enabled platform company: Will Saab disclose double-digit software revenue with margin expansion to low-teens EBIT by 2027?
This question breaks into three testable gates over the next 14 months.
The first gate is the February 2026 Capital Markets Day software disclosure. If Johansson quantifies software at greater than 10 % of revenue with margin profiles in the twenties, the platform thesis is validated. The stock likely re-rates 25 to 30 % as growth investors recognize the margin expansion potential and shorts covering. If disclosure shows 7 % or less, the narrative that software-defined is more marketing than substance gains credibility and the stock probably trades down 15 to 20 % toward traditional defense contractor multiples.
The second gate is 2026 execution evidence visible by mid-year. The backlog phasing concentrates delivery pressure exactly when new facilities are ramping. More than SEK 120 billion must convert to revenue in 2026 and 2027 while Saab simultaneously brings online factories in Michigan and India, integrates 2,700 net new employees hired in nine months, and scales T-7 trainer production that’s currently losing money. By Q2 2026 we’ll see whether operating cash flow has turned durably positive, whether Aeronautics margins are recovering as T-7 stabilizes, and whether the new facilities are ramping on schedule. If all three happen, the execution risk premium currently compressing the multiple releases. If any one fails, operational complexity becomes the dominant concern.
The third gate is megadeal conversion by year-end 2026. After 2027, backlog visibility thins materially with only SEK 32 billion scheduled for 2028 and SEK 26 billion for 2029 and beyond compared to SEK 63 billion for 2026 and SEK 58 billion for 2027. Saab needs at least one of the following: Ukraine financing solved and contract signed, two to three GlobalEye orders from European customers beyond France, or a major submarine export win. Without one of these, the post-2027 growth trajectory lacks support. With one or more, the revenue pipeline extends through 2030 and justifies continued platform investment.
There’s one risk almost no one is properly weighing. In the Q3 interim report risk section, Saab explicitly disclosed that rare earth elements are strategically important and that China controls significant global production. The disclosure continues that Saab is pursuing strategic stockpiling, alternative supply routes, and supplier diversification. This is unusually specific language for a risk disclosure. Most companies bury supply chain risks in boilerplate. Saab is calling out a specific material constraint and describing active mitigation.
China controls approximately 70 % of rare earth production and 87 % of processing capacity. Missiles, radars, electronic warfare equipment, and precision-guided munitions all require neodymium, dysprosium, and terbium. Lead time to develop Western alternative processing is five to seven years. If China restricts exports by 2027 or 2028, Western defense rearmament hits a binding physical constraint. The fact that Saab is already stockpiling and developing alternatives suggests they see this coming. If they’re ahead of competitors in securing non-Chinese supply, they gain monopoly-like pricing power during the constraint period. The appointment of Marcus Wandt to the new Strategy and Technology role immediately after the Ukraine letter of intent signals this is a top priority.
Three Scenarios and What to Do
At SEK 538, Saab trades at 41 times forward 2026 earnings and 23 times forward 2026 EV/EBITDA. The market is pricing something between a high-quality defense compounder and a platform company but hasn’t committed to either view. The next 14 months force a decision.
I see three scenarios with different probabilities and outcomes.
The bull case, which I assign 30 % probability, is that software disclosure exceeds 12 % of revenue with 25 to 30 % margins. Ukraine contracts for 60 to 100 aircraft. Multiple GlobalEye orders materialize and NATO standardization on Saab’s command and control architecture accelerates. Group EBIT margins reach 11 to 12 %. The stock re-rates to SEK 900 to 1,000, representing 70 % upside, as investors recognize platform economics. This warrants 30 to 35 times EV/EBITDA, which sounds expensive until you compare it to other software-enabled infrastructure providers.
The base case at 55 % probability is that software gets disclosed at 8 to 10 % of revenue with solid margins. Execution proceeds smoothly with cash flow inflecting positive and margins expanding to 10 to 11 %. Ukraine doesn’t materialize but GlobalEye wins two or three European customers. The stock trades to SEK 700 to 750 by end of 2026, representing 30 % upside and a 25 times EV/EBITDA multiple. This is a high-quality defense compounder with good growth but not full platform recognition.
The bear case at 15 % probability is that software disclosure disappoints below 7 %, revealing the platform narrative was more positioning than substance. Capacity ramp stumbles with facility delays or supply chain friction. Cash flow stays negative through 2026. T-7 losses worsen and drag Aeronautics margins. The stock reverts to traditional defense contractor valuation around SEK 450 to 500, representing 15 to 20 % downside and 18 to 20 times EV/EBITDA.
The expected value of these three scenarios weighted by probability is approximately SEK 730, or 36 % upside from current levels. That’s attractive but not compelling enough to significantly increase position size before we get more evidence.
My position is to hold at current levels. The evidence from Q3 strengthened the platform thesis, particularly the Arexis order quantifying software economics, but simultaneously revealed the magnitude of execution risk in 2026 and 2027. For patient capital with a three to five year horizon, current levels offer reasonable risk-adjusted returns. The expected value calculation supports holding, and the downside scenario requires multiple failures while the upside scenarios only require one or two things going right.
The honest assessment is this: I’m 60 % convinced the platform transformation is real and material, and 40 % concerned that operational complexity in 2026 could derail near-term financial performance even if the strategic direction is correct. Q3 was the quarter that moved the thesis from theoretical to falsifiable. We now have specific metrics to track, specific timelines to monitor, and specific proof points that will validate or invalidate each component.
The caterpillar versus butterfly metaphor is incomplete. We’re not debating whether transformation will happen. We’re watching the chrysalis in real time with partial visibility into what’s forming inside. Some structures are clearly emerging: the software architecture, the network effects, the NATO embedding. Other parts remain uncertain: the cash flow profile, the margin trajectory, the execution capability at scale.
February 2026 is when the chrysalis cracks open and we see what emerged. Until then, monitor the proof points religiously. The butterfly thesis is now testable. Q3 gave us the Arexis proof point that makes the valuation question answerable. The next 14 months tell us if we’re paying 40 times earnings for a platform company in its early scaling phase or an expensive caterpillar that never grows wings.
Disclaimer:
The content does not constitute any kind of investment or financial advice. Kindly reach out to your advisor for any investment-related advice. Please refer to the tab “Legal | Disclaimer” to read the complete disclaimer.