$MRLN: 54x Upside Hiding in a De-SPAC
Why every Collaborative Combat Aircraft will run on Merlin Labs' certified flight control layer, and why the market is pricing a $200B+ labor TAM at $257M of enterprise value
Shawarma Capital · April 2026
There is a company on the NASDAQ trading at $6.58 with an enterprise value of roughly $257 million that has, in my view, the cleanest path to becoming the operating system of record for every autonomous aircraft the US Air Force will field this decade. Its ticker is $MRLN. The market is pricing it like a $32 million revenue defense subcontractor with SPAC stigma. I think it is something else entirely. This essay is the case for why.
I am going to walk through this in five movements. First, the architectural thesis, what Merlin actually is inside the future combat aviation stack the Air Force has been quietly engineering for two years. Second, the Sequoia connection, why Julien Bek’s “Services: The New Software” framework, published March 5 of this year, is essentially a back-tested template for a company Matt George has been building since 2018. Third, the math, Crossroads Capital’s bear/base/bull scenarios and what they imply at the current price. Fourth, the receipts, the institutional footprints, the contracts, the certifications, and the insider behavior that tell you who has actually done the work. Fifth, the risks that could break the thesis, and the asymmetry that remains even if some of them play out.
This is research synthesis, not investment advice. I am long the equity. Do your own work. The disclosure is at the bottom. Let’s go.
I. The Stack Nobody’s Pricing In
In October 2025, Anduril’s YFQ-44A “Fury” went from a clean-sheet design to its first semi-autonomous flight in 556 days. In February 2026, the same aircraft flew with both Anduril’s Lattice mission autonomy and Shield AI’s Hivemind on a single sortie, swapping between them mid-flight. The same month, General Atomics’ YFQ-42A, officially nicknamed “Dark Merlin”, flew with Collins Aerospace’s Sidekick autonomy. The Air Force called this validation of its modular acquisition strategy. Most defense reporters covered it as a milestone for Anduril and Shield AI. They missed the deeper story.
What the Air Force is actually building is a four-layer stack. At the top sits mission autonomy, the tactical brain. This is what tells the aircraft what to do: target prioritization, swarm coordination, formation flying, sensor fusion, threat response. Anduril’s Lattice, Shield AI’s Hivemind, and Collins Sidekick all live here. They are tactical commanders made of code.
Below mission autonomy sits the A-GRA, the Autonomy Government Reference Architecture. This is the government-owned, open interface standard that lets any mission autonomy from any vendor talk to any airframe from any vendor. The Air Force published it explicitly to prevent vendor lock. Colonel Timothy Helfrich, the USAF portfolio acquisition executive for fighters and advanced aircraft, said the quiet part out loud in February: “We are not locked into a single solution or a single vendor. We are instead building a competitive ecosystem where the best algorithms can be deployed rapidly to the warfighter on any A-GRA compliant platform, regardless of the vendor providing the algorithm.”
Below the A-GRA sits Layer 3, flight control and contingency management. This is the layer that actually flies the aircraft safely. Stick-and-rudder control, fault detection, recovery from anomalies, certified airworthiness, real-time mission re-planning when something breaks. It is the dull, hard, life-or-death layer that has to be DO-178C Level A certified or military airworthiness approved before the plane is allowed to leave the ground. This is where Merlin lives.
Below Layer 3 sits the airframe itself, Anduril’s YFQ-44A Fury, GA’s YFQ-42A Dark Merlin, the MQ-20 Avenger, future production CCAs.
Currently, each airframe vendor has their own homegrown Layer 3 stitched into their aircraft. Anduril rolled their own. General Atomics has spent years refining theirs on the MQ-20. The mission autonomy companies plug into these proprietary stacks via A-GRA. The February 2026 demonstration validated that architecture. But the deeper architectural question, and the question Air Force planners have already started answering, is whether the bottom layer should also be modular and competitively bid, with a certified, vendor-neutral substrate underneath every airframe.
Merlin already signed the contract for that role. In October 2025, Merlin signed a Cooperative Research and Development Agreement (CRADA) with the USAF specifically to build A-GRA contingency management capabilities, autonomous fault detection, recovery, and real-time mission adaptation. Major Dustin Graves at AFWERX framed it directly: “This CRADA will help us build on a common, government-owned architecture for autonomous systems, ensuring interoperability, accelerating innovation, and ultimately, delivering resilient and adaptable capabilities to our warfighters.”
That is Layer 3. That is the brainstem. And Merlin is the only public-market company that has the certification pedigree, the cultural alignment, and the Tier-1 partnerships to fill the slot.
Anduril and Shield AI cannot. Anduril went from clean-sheet to first flight in 556 days, which is genuinely incredible for an airframe but is the philosophical opposite of how DO-178C Level A flight control software gets built. DO-178C Level A typically takes three to five years and requires every line of code to be traceable to a documented requirement, every requirement to be traceable to a verification test, and every test to pass before any change is accepted. Anduril and Shield AI’s competitive advantage is rapid iteration, software-first deployment, and tactical innovation cycles measured in weeks. Their entire culture is the opposite of certification work. They could spin up a certified flight control practice — but they would burn three years and $200-400M doing it, against a competitor who has a five-year head start.
Merlin’s Layer 3 thesis is additive to everything Crossroads modeled. The $1.6B ARR base case Crossroads built was for retrofit on legacy fixed-wing, C-130J, KC-135, Cessna Caravan, future commercial cargo. None of it counted CCAs. The Air Force has stated a goal of fielding at least 1,000 CCAs by 2030. Each is a new-build aircraft that needs certified Layer 3 flight control. Per-tail economics likely look similar to legacy retrofit math, call it $2-3M integration plus $1-2M annual license. At 1,000 CCAs that is $1.5-4B of additional ARR opportunity stacked on top of the existing thesis. And that is just the US program. Allied CCA programs in Japan, Australia, the UK, and others extend the architecture further. If A-GRA becomes a NATO/Five Eyes interoperability standard and Merlin is the certified Layer 3 reference implementation, the international ramp adds another multi-billion-dollar leg to the same thesis.
The market is pricing none of this.
II. Matt George Said It First — Five Years Before Bek
On March 5, 2026, Sequoia Capital partner Julien Bek published an essay called “Services: The New Software.” The thesis: the next trillion-dollar company will not sell software tools. It will sell the work itself. Bek’s argument has six load-bearing claims.
First, every dollar spent on software is matched by six dollars spent on services, so the labor budget is six times the size of the software budget, and that is where the real opportunity lives. Second, intelligence work (rule-following) and judgment work (taste built on experience) are different categories, and AI is now good enough to automate most intelligence work autonomously. Third, the higher the intelligence ratio in a profession, the sooner autopilots win. Fourth, the right wedge is wherever the work is already outsourced, because then it is a vendor swap rather than a reorg. Fifth, today’s judgment becomes tomorrow’s intelligence as the autopilot accumulates proprietary data. Sixth, and this is the line that should make every existing aviation incumbent uncomfortable, in 2026, copilot companies will try to become autopilots and run into the innovator’s dilemma: selling the work means cutting their own customers out of doing it. That is the opening for pure-play autopilots.
Merlin checks every single box. Not approximately. Literally.
The most striking thing buried in a May 2021 TechCrunch interview is Matt George’s exact words to reporter Aria Alamalhodaei: Merlin Labs has no intention of becoming an airline or operating planes themselves. Instead, it is looking to provide autonomy as a service. He used the phrase in 2021. Bek published “Services: The New Software” in March 2026, four years and ten months later. George was not responding to Bek’s framework. He had built his entire company architecture around it before Bek had named it.
George’s March 2026 shareholder letter goes further. He writes that Merlin is “doing the really hard work of taking everything that’s happened in a pilot’s brain and putting that into certified software to enable an aircraft to think on its own.” Compare directly to Bek: “Today’s judgment will become tomorrow’s intelligence. As AI systems accumulate proprietary data about what good judgment looks like in their domain, the frontier will shift.” George is not just building autopilot software. He is building the judgment-to-intelligence conversion machine for aviation.
Then George closes the loop himself: “Every aircraft we equip will generate data that makes the system smarter. Every flight expands our training set. The more we fly, the more valuable the software becomes, and the harder it becomes for anyone else to replicate what we have built.” That is Bek’s data flywheel, executed in defense aviation, written months before Bek’s essay went live.
There is a deeper reason Merlin is the cleanest fit for the autopilot framework Bek describes. Aviation is the only profession in human history that has spent 100 years systematically converting judgment into rules. Every fatal accident is investigated by the NTSB, the findings codified into Federal Aviation Regulations, the regulations baked into checklists, the checklists tested in simulator recurrents. When Sully landed on the Hudson, his judgment became a checklist revision. When the 737 MAX MCAS killed 346 people, Boeing’s failure became software changes. The phrase “written in blood” comes from aviation. There is no other profession where tacit expertise has been so aggressively transformed into explicit, machine-readable rules, for an entire century, by regulatory mandate. The Merlin Pilot listens to ATC radio (NLP, pure intelligence task), interprets the instruction against published procedures (rule-matching, pure intelligence), checks the configuration against the checklist (verification, pure intelligence), executes the maneuver via control surfaces (deterministic actuation, pure intelligence). The “judgment” portion is dramatically smaller than in legal work or medicine because aviation has been pre-converting judgment into rules for a century. Merlin doesn’t have to invent the judgment-capture layer that critics of the Bek thesis say is the hard part. The FAA already did it.
The outsourcing wedge is also already there, and almost nobody outside the industry knows it. The conventional read is that military pilot operations are insourced. The conventional read is wrong, and has been for at least a decade. The DoD has been quietly outsourcing huge chunks of flight operations under Contractor-Owned, Contractor-Operated (COCO) services. Metrea Strategic Mobility, a private company, operates fifteen KC-135 tankers under contract to the USAF. The Air Force is literally paying private contractor pilots, in private contractor KC-135s, to refuel its own aircraft. CAE USA has held the KC-135 Aircrew Training System contract since the early 1990s. Pilot training has been outsourced for over thirty years. Pilot operations have been outsourced for over a decade. Merlin is not displacing uniformed crew, it is positioning to displace the contracted crew slot, the slot the Air Force has already accepted as outsource-able.
Now run Bek’s $6:$1 ratio on flying. The “tool” budget for aviation autonomy software today is essentially zero. The “work” budget is staggering. Labor costs surpassed fuel as the U.S. airline industry’s largest operating expense in 2024 at 36.8% of total OPEX. Pilot contract settlements between 2023 and 2025 delivered 30-50% cumulative pay increases at all five majors. Widebody captains now earn $450,000-$500,000 annually. Boeing’s Pilot and Technician Outlook estimates the world will need 660,000 new commercial airline pilots over the next 20 years. A 24,000-pilot shortfall is projected to peak in 2026, driven by 4,300 mandatory retirements per year through 2042 and the 1,500-hour ATP rule constraining new supply. The full labor TAM Merlin is addressing isn’t the autonomy software market, it is the global pilot wage bill plus military aviator lifecycle cost plus contracted COCO operations. That is somewhere between $200B and $400B annually, growing 5-7% per year for the next two decades.
And the innovator’s dilemma is already playing out, but in Merlin’s favor. Honeywell, Northrop Grumman, and GE Aerospace each signed formal partnership agreements with Merlin in late 2024 and 2025. Honeywell’s October 2024 MOU integrates Merlin’s solutions with its Aerospace Technologies division. Northrop’s June 2025 agreement uses Merlin as the platform to validate mission-autonomy software through its Beacon testbed. GE Aerospace’s September 2025 “Autonomy Core” Program of Record links Merlin directly to GE’s Flight Management System, the operating system of record for over 14,000 aircraft globally. Three of the largest copilot incumbents in the entire aviation ecosystem looked at the autopilot transition, looked at their own innovator’s dilemma, and decided they couldn’t build it from inside their existing business models. So they outsourced the autopilot layer to Merlin. Honeywell isn’t going to build software that tells airlines they don’t need pilots, but it’ll integrate with Merlin’s software that does, because at least then Honeywell stays in the cockpit.
This is the precise pattern Bek predicted: pure-play autopilots winning the autonomous layer because incumbents can’t bring themselves to attack their own customers. Merlin won the GE FMS slot specifically because GE couldn’t build the autopilot itself without breaking its relationship with Boeing/Airbus pilots.
George just didn’t have the vocabulary in 2021. Bek gave it to him in 2026.
III. The Crossroads Math
On March 23, 2026, Crossroads Capital published a public investment thesis on Merlin, written by Ryan O’Connor, with an entry at $6.80 per share via BACQR rights. The full report runs to a comprehensive bear/base/bull framework. I am going to reproduce the price targets directly because they are the cleanest articulation of the asymmetry I have seen anywhere in defense tech right now.
Read those numbers slowly. The bear case is $53 per share, eight times the current price. That is the scenario where Merlin only converts 300 aircraft of its current contracted fleet and trades at a humble 7.5x revenue multiple as a defense services contractor. The base case is $366, fifty-four times the current price, and assumes Merlin executes against the 800 aircraft already under its contract ceiling at a 20x EV/Revenue software multiple. The bull case is $1,708, over 250x, and assumes Merlin captures roughly 6,000 aircraft globally including allied fleets and commercial cargo, valued as a pure software platform.
The base case requires Merlin to do exactly what its existing contracts already authorize. The $105M USSOCOM IDIQ for the C-130J, signed June 2024 under SBIR sole-source authority, has a contract ceiling that supports the path to 300 aircraft on its own. The $16M USAF KC-135 program at MacDill AFB targets a 376-aircraft fleet. Together that is 676 aircraft already under contract, before the GE Aerospace Program of Record opens access to 14,000+ commercial airframes through the FMS integration.
The unit economics make this tractable. Per Crossroads: each integration is a one-time $3M fee, each aircraft generates $2M in annual recurring software license, each aircraft has a 10-15 year operational life, customer payback is under one year (because pilot labor savings run roughly $7M annually per aircraft once you go to single-pilot or uncrewed), and customer ROI is 3-4x annually. The economics close cleanly for the customer, which is why the contract pipeline is what it is. This is not a science project. The Merlin Pilot has logged hundreds of autonomous flights across at least seven aircraft types: C-130J Super Hercules, KC-135 Stratotanker, Cessna 208B Caravan, Beechcraft King Air, de Havilland Twin Otter, Rutan Long-EZ, and Cozy Mark IV. From small experimental airframes to large military transports. Aircraft-agnostic architecture isn’t marketing.
And the brainstem thesis is additive to all of it. Crossroads modeled retrofit. They did not model Merlin becoming the certified Layer 3 substrate underneath every CCA program globally. The ~$1.5-4B of additional ARR I estimated in Section I sits entirely on top of the bear/base/bull scenarios above. If A-GRA becomes the standard, the bull case is potentially conservative, not aspirational.
The market is currently pricing Merlin at roughly $0.54B EV. By comparison: Anduril is at $30.5B private, Shield AI at $12.7B private, Joby at $7.8B public, Aurora Innovation at $13.2B public, Rocket Lab at $12.5B public. Merlin trades at roughly 1/56th of Anduril and 1/24th of Shield AI. None of those companies has a public-market FAA certification basis. Merlin does, and is the only autonomy company in this peer set that does.
IV. The Receipts
The most important tell in any de-SPAC is who actually showed up and stayed. Here are the people who put real money into Merlin and what their behavior tells you.
Baillie Gifford, the firm that backed Tesla at $6 per share, held Amazon through a 30x return, and was an early investor in SpaceX, was a Merlin investor before the SPAC. They increased their position at every subsequent round, including the PIPE upsizing from $125M to $200M+. Baillie Gifford’s pattern is to identify category-defining technology companies at inflection points and concentrate. Their increased participation through every round is among the cleanest institutional signals I have seen on a microcap.
Citron Research- historically a short-seller, published a public note in October 2025 calling BACQ “the most compelling risk/reward stock in the market today” and set a $30 price target. Citron compared Merlin’s $800M valuation on real defense contracts against IonQ’s $23B on largely theoretical quantum computing. BACQ surged 12% on the endorsement. When a short-seller goes long publicly, that is one of the strongest conviction signals in public markets. Citron’s whole career is built on identifying companies that sell narrative tools versus companies that sell measurable outcomes. They short the narrative companies. They go long the outcome companies. Merlin is, in their mental model, an outcome company.
Michael Blitzer- the SPAC architect, runs Kingstown Capital, a multi-billion AUM hedge fund. He trained under Joel Greenblatt at Gotham Capital. His prior SPACs include Intuitive Machines (LUNR, +65% from trust) and USA Rare Earth (USAR, +41%). He called Merlin “a national asset.” His 8.80M sponsor shares are locked up with economics tied entirely to long-term performance.
The 100% equity rollover. Every single legacy Merlin shareholder rolled 100% of their equity into the public company. No partial liquidity at close. Matt George personally has somewhere between $80M and $240M of net worth riding on the same shares as public investors. No insider sold a single share.
The 90.3% redemption. When SPAC shareholders had the chance to redeem their trust shares at $10, 90.3% of them did. Crossroads framed this as deliberate cap-table engineering, insiders accepting a low IPO price to accumulate before price discovery began. I do not buy that read. SPAC investors are a distinct class. They buy SPAC units, BACQU in this case, which decompose into a common share (BACQ) and a right (BACQR). The cash sits in trust earning interest. At de-SPAC, the arbitrageur redeems the common at $10, pockets the accrued interest, and keeps the right to sell in the secondary market. It is structured short-term carry, not a long-term equity view. Look at the pre-deal holder list. Almost none of them are long-duration growth investors. The 90.3% redemption tells you nothing about Merlin’s quality. It tells you the SPAC arb crowd took their free money and left, exactly as designed.
The follow-on math also undercuts the engineered-low-price theory. If insiders had orchestrated a thin float to compound their own accumulation, they would not have turned around and raised roughly $80M in a dilutive offering shortly after the de-SPAC. They did. Merlin is in early growth and needs cash more than it needs a clean cap table, and the subsequent raise confirmed that priority. The thin float is real. The disproportionate price action on any catalyst is real. The story about why the float is thin is wrong.
The CMO hire on March 31. Merlin appointed Michael Baker as Chief Marketing Officer on the day the stock printed its low. Baker brings 15 years of brand and communications experience from Formlabs, where he was Global Head of Brand and Communications. The hire aligns precisely with what Crossroads predicted: “When the cadence of promotion begins in the coming weeks, we doubt the repricing will be gradual.” The CMO appointment marks the start of the active promotion phase. It is the inflection point where Merlin transitions from quiet accumulation into telling its story.
Zero analyst coverage. As of right now, no Wall Street sell-side analyst covers MRLN. Zero. The discovery phase has not even begun. TD Cowen is the most likely first mover.
Seven facilities, seven aircraft types, 200 employees, 33 open roles. Merlin operates from Boston (HQ), Denver, Quonset State Airport in Rhode Island (the certification flight test center), Mojave (the original stealth-mode test site), Kerikeri New Zealand (where two Cessna Caravans run flight testing under the NZ CAA SOI-2 certification track), MacDill AFB Florida (KC-135 data collection with the 6th Air Refueling Wing), and a new flight test hub at Hanscom Field Massachusetts opening in early 2027. Merlin’s open roles include a Senior DO-178C Flight Software Engineer, DO-178C is the FAA certification standard. You do not hire DO-178C engineers unless you are actively pushing certification milestones.
The MDA SHIELD / Golden Dome connection. Merlin was awarded a spot on the $151 billion SHIELD IDIQ in late 2025, the primary contract vehicle for Trump’s Golden Dome missile defense initiative. This is separate from and additive to the C-130J and KC-135 contracts. The FY2026 defense budget allocates $13.4 billion specifically for autonomy as a brand new standalone budget line. This is the structural tailwind that makes the Bek labor TAM realistic.
V. The Asymmetry — and the Risks
Let me be honest about what could break this thesis, because some of it might.
The going concern flag. On January 13, 2026, the auditor raised “substantial doubt about the Company’s ability to continue as a going concern” in the pre-SPAC financials. This was almost certainly resolved by the $200M+ in PIPE proceeds, but the language is in the filing and short-sellers can weaponize it.
The 12% preferred coupon. The Series A Preferred shares carry a 12% annual coupon on 10.2 million shares. That is roughly $12.3 million per year in mandatory preferred dividends, on top of operating cash burn. Combined with the $61.6M projected operating burn for 2026, real annual cash outflow is closer to $74M. At $146M of cash on the balance sheet, the realistic runway is closer to 2 years, not 3-4. Merlin may need to raise again before the production contracts convert.
Revenue actually declined during the defense pivot. Revenue went from approximately $10M in 2022 to $8.5M in 2025E. The 2026 ramp to $32M is the inflection, but it has not happened yet. Net loss for the nine months ending September 2025 was $55.73M, up from $41.83M in the prior nine-month period. Losses are widening, not narrowing.
Matt George’s prior startup, Bridj, failed. George founded Bridj in 2014, an on-demand shuttle bus startup. He raised $11M, hired 50 employees, and the company died in 2017 when Toyota pulled an investment at the last minute. He was 26. The bull read is that he learned from failure, then built Merlin for seven years before going public, and secured defense partners first. The bear read is that Glassdoor reviews call him a “narcissist” and complain of disorganization, shifting priorities, and absent management. There are eight Glassdoor reviews. They average 3.6/5. The CEO is divisive. This is real.
SBIR eligibility risk. The $105M USSOCOM IDIQ was awarded under SBIR sole-source authority, which requires Merlin to qualify as a small business under SBA size standards. As Merlin’s valuation grows, it may eventually exceed those standards and lose eligibility for future SBIR awards. Current contracts are grandfathered, but new ones could be at risk. Growing the company may close the door on the mechanism that awarded its biggest contract. Almost nobody is discussing this.
The Anduril vertical integration risk. Anduril is famously vertically integrated. Their cultural distrust of outside vendors is deep. If Anduril decides to absorb the certification burden and build Layer 3 in-house, which they could, at the cost of three years and several hundred million dollars, they cut Merlin out of the Fury program entirely. Same logic applies to Shield AI on V-BAT and other platforms. The brainstem thesis works only if the Air Force forces both companies to outsource the certified substrate, and the Air Force has not yet made that final call. The CRADAs Merlin and Reliable Robotics signed are evidence the Air Force is leaning that way. They are not yet a guarantee.
Reliable Robotics is the head-to-head competitor for Layer 3. Reliable also has a CRADA. Reliable also has FAA cert pedigree. Reliable also has a Cessna Caravan platform. If the Air Force picks one Layer 3 substrate vendor, it could be Reliable instead of Merlin. The fact that both are CRADA partners suggests the Air Force has not picked a winner, and may never pick a single one.
SPAC stigma. De-SPACs have terrible track records as a category. Many institutional investors have blanket policies against SPAC-originated equities regardless of fundamentals. The 61.6% day-one drop reflects this dynamic, not Merlin’s business quality. The stigma is real and may persist longer than the fundamentals justify.
Now stack the asymmetry against the risks. At $6.58, you risk roughly 1x, your entire position, if Merlin fails to execute, runs out of cash, and gets diluted to zero. The Crossroads bear case at $53 is a 7.7x return, achieved by Merlin merely executing on 300 aircraft of its existing contract base. The base case at $366 is a 54x return. The bull case at $1,708 is a 251x return. You risk one to make seven in the bear case, fifty in the base case, two hundred and fifty in the bull case. Even if you put a 70% probability on the bear case, a 25% probability on the base case, and a 5% probability on the bull case, the expected return is somewhere between 100x and 200x. That is the mathematical definition of asymmetric.
This fits a specific kind of opportunity: a category-defining company that entered the public market through a channel the market has learned to reflexively dismiss, at a moment when the discovery phase has not begun, with the float collapsed by SPAC arb redemptions rather than by any fundamental verdict on the business, with the institutional smart money already concentrated, with a path to $200B+ of labor TAM that nobody is yet pricing in, in a profession that has spent a century pre-converting judgment into rules, validated by the three largest aviation incumbents who chose Merlin specifically because they couldn’t build it themselves.
The CMO hire on March 31 was the starting gun for the promotional phase. The first analyst initiation will almost certainly come in Q2 or Q3 2026. The first earnings report will establish the growth narrative with institutional investors. The C-130J production contract conversion is expected Q2-Q3 2026. KC-135 program expansion is expected Q2-Q4 2026. FAA SOI-3 and the eventual STC issuance are 2027-2028 events, and represent the truly transformative moment, the first certified AI skill on a manned aircraft.
The thesis pays out as those catalysts land. None of it is in any sell-side model because there is no sell-side coverage. Zero.
At $6.58, the market is pricing Merlin like a $32 million revenue defense subcontractor with SPAC stigma. I think it is the certified flight control substrate every Collaborative Combat Aircraft will plug into via A-GRA, the autopilot Bek was describing for the profession that has spent a century pre-converting judgment into intelligence, the operating system of record for high-assurance aerial systems globally, and one of maybe five or six companies in the world built from day one to be the answer in its category.
The market will figure it out. The only question is whether it does so before or after the catalysts force it to.
Disclosure & Methodology
I am long $MRLN equity. This is research synthesis, not investment advice. You should not buy or sell securities based on anything I write. I am not a registered investment advisor; I do not owe you a fiduciary duty; my conclusions could be wrong in ways I have not anticipated. Do your own due diligence.
Sources for this essay include: SEC filings (8-K, S-4, S-4/A); the Crossroads Capital public investment report by Ryan O’Connor (March 23, 2026); Citron Research’s October 2025 BACQ note; Julien Bek’s “Services: The New Software” (Sequoia Capital, March 5, 2026); USAF press releases and AFWERX announcements regarding A-GRA; Anduril Industries press releases on YFQ-44A; Shield AI press releases on Hivemind integrations; FlyingMag, Breaking Defense, DefenseScoop, AeroTime, The Aviationist, Joint Forces News, ExecutiveGov, and Aerospace Global News reporting on the CCA program; CSIS analysis on MOSA and burden-sharing; Boeing Pilot and Technician Outlook 2024; Matt George’s March 2026 shareholder letter; multiple TechCrunch interviews dating to 2021; Boston Globe coverage of Bridj; USPTO patent filings; Glassdoor reviews; GE Aerospace, Honeywell, and Northrop Grumman partnership announcements; the company’s own published materials; LinkedIn employee data; and approximately 100 additional sources I have catalogued across this research stream. Where I have made my own estimates — particularly around the additive CCA Layer 3 TAM — I have flagged them as such.
The price target framework is Crossroads Capital’s. The brainstem thesis and the connection to Bek’s framework are mine.
If you found this useful, subscribe to Shawarma Capital. If you think I am wrong about something specific, push back in the comments — I would rather find out I am wrong now than later.
Shawarma Capital writes about asymmetric opportunities at the intersection of hardware, defense, and semiconductors. This is the deepest piece I have published.
this one aged pretty well
Really well done. Will increase my position after reading.