Crossroads Capital Q1 2026 Investor Letter

(MENAFN- ValueWalk) Crossroads Capital's commentary for the first quarter ended March 31, 2026.

Read more hedge fund letters here

During the first quarter of 2026, Crossroads Capital Investment Partners, LP (the“Fund”) appreciated by 4.2% net of all fees and expenses. On a five-year annualized basis, the Fund has returned 8.6% gross and 7.0% net. Since inception, it has compounded at a rate of 21.3% gross and 17.1% net.

As of March-end 2026, the Fund's total non-delta-adjusted gross and net exposure stood at 114.1% and 73.3%. Additional information on both the current portfolio and our historical gross and net returns is available at

Q1 2026 Market Commentary Update

In our Q4 2025 update, we highlighted a market that had metabolized 2025's uncertainty with surprising composure, once an understanding of the behavioral patterns in some key individuals was better internalized. Q1 put that cognitive capacity to a real test. Geopolitical escalation tied to Trump's increasingly unpredictable foreign policy produced the kind of quarter in which a manager's judgment in the face of uncertainty is put on display-for better or worse.

The S&P 500 declined approximately 4.3% in Q1. Standing alone, that number looks disappointing. But consider that Brent crude surged more than 90% over the same period, driven by the Iran conflict and the effective closure of the Strait of Hormuz-a supply shock that, in any prior cycle, would have invited a far more punitive repricing of risk assets. That it didn't is worth sitting with for a moment. As a humorous side note, we'd only ask“Wait, what about Greenland? Did we move on?” Frankly, yes. We did.

At their core, markets are complex adaptive systems. The actors within them-Crossroads, allocators, traders, algorithms, pension funds-don't respond to the same stimulus the same way twice, as each experience rewires the network's collective decision architecture. This phenomenon is precisely what played out in Q1.

In early 2025, Trump's tariff escalation triggered a sharp, indiscriminate selloff. The system had no recent memory of that input, so participants did what participants do when facing a novel threat: they sold first and asked questions later. But that selloff was followed by an equally sharp reversal that left those who panicked stranded on the wrong side of the trade, a scar now etched into the market's muscle memory. Fast forward to Q1 2026 and you have a fundamentally more severe shock, a 90% move in Brent crude, an actual shooting war, a critical chokepoint for 20% of global oil supply physically shut, and counterparties actively engaging antagonistically with the market to induce a response in their key adversary. Yet the equity market's aggregate response was notably more muted-not because participants lacked fear, but because the system had adapted. The same actors who sold into the tariff chaos learned the hard way that Trump-driven dislocations tend to reverse, and that missing the turn carries a steeper cost than riding the drawdown.

In Q1, therefore, the market was essentially in a state of suspended animation. Average investors, awaiting the President's next Truth Social post, diplomatic reversal, or naval maneuver, found themselves unable to commit in either direction. They were concerned about potential downside, but didn't want to miss out on potential upside. We believe Trump (along with key members of his administration) has been leaning into this phenomenon, becoming the market's puppet master: every escalation followed by a hint of de-escalation, every stick by a carrot. He's keeping a critical mass of capital frozen in a wait-and-see posture that resembles last year's tariff paralysis but is decidedly its own animal, suppressing the kind of cascading liquidation you'd otherwise expect from a 90% oil shock.

Some might call this complacency, and argue that liquidation is just around the corner. We'd push back. We're observing a market that has internalized risk but refuses to capitulate to it-and for reasons that are actually quite rational. Last year we argued that tariffs themselves weren't the market's concern; uncertainty was. This year, the script has flipped: The oil shock is real, but uncertainty is far diminished. We know how this President operates. His adversaries in the Middle East may not be reliable trade partners, to say the least, but the current scenario there has a foreseeable set of outcomes with less variance than last year's big event. And if we had to guess, once Iran is in the rearview, we'll be moving on to Cuba, a geopolitical sideshow with far less economic consequence. Again, we'd ask“Hey, remember Greenland?”

As students of complex adaptive systems, we'd add that this dynamic carries its own risks. A network that has learned to buy the dip can, under the right conditions, learn that lesson too well. It can suppress volatility right up until some exogenous input exceeds the system's adaptive capacity. At that point, the repricing is violent precisely because everyone was positioned for the bounce. We are not predicting that outcome, but we are aware of it-and as always, we're prepared to act should it rear its ugly head. Regardless, we remain anchored to our north star: building a portfolio around businesses with identifiable value-unlocking transformations and hard catalysts, not macro bets.

The oil market told a similar story of complex adaptation. The closure of Hormuz didn't produce the single-variable catastrophe a textbook would predict, because the system re-routed around it. China pulled forward refinery maintenance, curtailed product exports, and leaned into Russian and non-Gulf barrels. The UAE walked out of OPEC in early May (and remains out as of this writing), removing one of the few members with real spare capacity and accelerating the cartel's structural erosion. The U.S, once again playing the role of supplier of last resort, pushed product exports to a record 8.2 million barrels per day as European and Asian refiners scrambled for anything that didn't transit the Gulf. On top of that, demand destruction is already here, and it is not evenly distributed: the IEA expects a 1.5 million barrel per day year-over-year decline in Q2 (the steepest since COVID) from India, Pakistan, Nepal, Bangladesh, Indonesia, Egypt, Ethiopia, and others. None of this resolves the shock. Rather, it redistributes it. Prices remain elevated, inventories are drawing down, and the system is running hot. But the catastrophic tail has, so far, been absorbed by a network of participants doing exactly what complex adaptive systems do: finding another way through.

Encouragingly, Q1's fundamentals were consistent with the broadening we've been waiting for. While the cap-weighted S&P 500 fell 4.3%, the equal-weight index and Russell 2000 each posted modest gains of roughly 0.7% and 0.9% respectively, extending the rotation away from mega-cap concentration we flagged in prior letters. The Magnificent 7 continued to underperform the broader market, and what we described as a long-awaited secular shift appears to be gathering structural momentum. For a strategy like ours, concentrated in smaller, under-followed businesses trading at deep discounts to their medium-term earnings power and supplemented by idiosyncratic special situations that lack correlation to broader markets, this backdrop could not be more encouraging.

Q1 2026 Performance Attribution Update

In a quarter where the S&P 500 declined 4.3% and the Magnificent 7 fared considerably worse, we were pleased to see the portfolio hold its ground on the whole, with special situations kicking into gear.

Our largest contributor was Merlin Labs (MRLN), contributing to 40% of our Q1 net gains, as it started trading as a public company in mid-March and we both closed out the short position on the common as our rights converted and took advantage of the early-days volatility. We detailed our thesis extensively in a published research report, but the early price action is consistent with what we expected: a“tiny public float, locked-up insiders,” and a market that hasn't done the work yet on a name that entered through the SPAC channel-the one channel institutional investors have been trained to ignore. We suspect the repricing has barely started, so the company is firmly in both our special situations bucket and our emerging compounder bucket if milestones are met.

AST SpaceMobile (ASTS) contributed 23% of our gains in Q1, continuing to reward patience as the constellation buildout progresses and the market slowly digests the implications of what this company has already demonstrated. FTAI Aviation contributed 17%, and Nebius (NBIS) contributed 14%; both are names where our original thesis remains intact and the businesses themselves continue to execute.

On the downside, Nintendo (NTDOY) was our largest detractor, comprising 35% of our losses, followed by Vistry (VTY) at 19% and Company X at 13%. On Nintendo, the market remains fixated on near-term memory prices and we continue to believe that investors are missing the forest for the trees. Vistry was a different story. We held a small position and exited during the quarter as the final nail in the coffin (CEO Fitzgerald resigning) confirmed that this transformation has meaningfully greater uncertainty than we'd originally underwritten. We'll revisit the name once the heavy lifting is behind the company, whenever that may be, but for now our capital is better deployed elsewhere.

We've said it before, but it bears repeating: The gap between being right and being rewarded is where most investors lose their nerve. Our portfolio is not built to outperform in any particular quarter. It's built to compound over years as the transformations we've identified play out on their own schedule-not the market's, and certainly not ours.

Quarterly Investment Activity Update

Several of our names carried their late-2025 momentum into the new year, and we saw little reason to get in the way. The core book was largely untouched in Q1. When businesses are compounding and the thesis is intact, the best thing a PM can do is stay out of his own way.

Faithful readers will recall that we hinted at the Merlin Labs special situation roughly six months ago. The trade began as a paired position: long BACQR rights and short BACQ common. We aimed to capture the spread while hedging the risk that the deal would fail to close. As the common broke toward the rights-implied price just ahead of the merger, we closed the short leg at a profit. We also trimmed the long side into strength, locking in gains on a portion of the position before letting the remaining rights convert through the de-SPAC. The arbitrage has run its course, but our involvement with Merlin has not. Our March report lays out our full thesis, and we're in this one for the long haul.

Next up is a quick comment on Nebius, which has been equal parts rewarding and humbling. When we initiated the position last year, the company had a lot to prove, though we saw a massive margin of safety and described the business as trading below liquidation value. We liked the team, we liked the assets, and we were right on the thesis. But if we're being honest with ourselves, we sized it too small. Our methodology is to start small and average up as milestones get achieved. In this case, the milestones came so fast that we should have matched that pace with our own, adding more aggressively as the business de-risked in real time as opposed to our more patient approach. First-world problems, admittedly, but when a company is executing at that tempo, our investment pace needs to keep up. We've internalized that lesson, and will act accordingly going forward. But that bump in the road didn't throw us off track: We stayed true to our methodology, and with time and performance, we now hold a more aggressive position in Nebius while the business continues to fire on all cylinders. Indeed, we think its best days still lie ahead.

We continue to evaluate several new names that we expect to discuss with partners in the coming quarters. Keep on the lookout for our emails on single stock ideas, which we share very rarely nowadays. As longtime readers know, we prefer to build positions quietly before talking about them publicly. More to come.

Notable Position Updates Nintendo (NTDOY)

In the last three months, two of the three pillars of the bear thesis we walked through last quarter quickly crumbled. Their collapse didn't vindicate the bull case outright, as memory pricing is still a live debate. But tariff impacts and the“no first-half system sellers” takes are off the table. We plan to publish a standalone piece on the memory cycle and the year ahead for the company, whose prospects we believe the market is discarding wholesale amid trend-following taken to its extreme.

AST SpaceMobile (ASTS)

Q1 picked up exactly where Q4 left off. The company's transition from R&D-stage startup to operational scaleup, which we described last quarter, went from“underway” to“unmistakable” over the course of the last three months. There was one setback, as BB7 was placed in the wrong orbit by the New Glenn 3 rocket, sparking a downturn that had everything to do with Blue Origin's vehicle misplacement, not any failure of AST's technology. Nonetheless, the setback served as a healthy reminder that navigating the space frontier is never without challenges, particularly for a mission of this scale.

The company's early March earnings update showed full-year 2025 revenue came in at $70.9M, at the top end of the guided range, driven by 15 commercial gateway deliveries across nine MNO customers on five continents and milestones against ten active government contracts. 2026 revenue guidance is $150–200M, at least a doubling, and management gave clarity and context to the $1.2B of contracted backlog and government-related scaling we should see into next year. Q1 2026 revenue of $14.7M was light relative to consensus, but guidance was reaffirmed and management noted revenue will be heavily weighted toward the second half of the year as launches begin and commercial service activates.

Pro forma liquidity stands near $3.5B, which is more than enough to fund the constellation buildout and get AST to commercial service without going back to the market. However, management said this was a raise to go“on offense,” and we're excited to see what materializes as the company goes from strength to strength. A year ago, the loudest short argument was dilution-to-death. That seems incongruous with today's fact pattern.

It's in manufacturing where the scaleup is revealed in hard numbers. AST is now in the advanced stages of production and assembly through BlueBird 33, with phased arrays completed through BlueBird 28. Manufacturing footprint exceeds 500,000 square feet, and the Texas Micron facility is operational, with capacity supporting over ten satellites per month. BlueBirds 8, 9, and 10 are scheduled for a mid-June Falcon 9 launch from Cape Canaveral, and we expect consistent shipping and launches from that point forward through the end of the year, targeting approximately 45 satellites in orbit by December.

Most importantly, Avellan confirmed on the call that single-satellite launches end with BB7: future missions will stack in groups of three, four, six, or eight satellites, which is the entire reason the New Glenn seven-meter fairing matters. AST's total number of satellites used to grow by small steps, but from now on it'll be giant leaps.

Which brings us to the new Block 2 bus. It uses a lighter composite structure to maximize payload efficiency, and we believe that required rework of the structure to maintain shape under launch loads and interfaces associated with stacked configurations on New Glenn are the cause for the delay in batch launches versus previous expectations. That's a meaningful engineering item, and from what we understand the company worked through it during Q1. The mid-June launch of BB8–10 should confirm that the problem has been resolved.

Everything else continues to compound. On the commercial side, the partner base now exceeds 60 MNOs covering more than 3 billion subscribers, and AST demonstrated a peak download speed of 98.9 Mbps to an unmodified off-the-shelf smartphone using Block 1 BlueBirds-a benchmark that validates commercial viability. Management also laid out the spectrum stack in more detail than we'd seen previously: approximately 1,150 MHz of tunable low- and mid-band MNO spectrum globally, 45 MHz of MSS lower mid-band in North America, and 60 MHz of licensed S-band priority rights outside North America. Avellan also telegraphed a mid-band constellation beginning to launch by the end of 2026, which is not in the model and meaningfully expands the TAM beyond the core D2D story. On the government front, AST layered a $30M prime contract from the Space Development Agency for HALO Europa Track 2 on top of the SHIELD IDIQ award from January, another direct-to-device national security workstream bolted onto the Golden Dome trajectory. Government revenue doesn't require the full constellation, scales linearly with satellite count, and for the highest-value direct-to-device contracts we believe AST remains the only bidder on Earth with demonstrated capability.“Competitive procurement” for the contracts that matter is still a formality. The FCC has also now authorized commercial SpaceMobile service in the U.S., a regulatory milestone that clears the path to revenue as satellites reach orbit.

Net-net, the company is executing and we believe it's past the manufacturing bottleneck, with the mid-June launch serving as the definitive proof point. The revenue engine is on, manufacturing is running to cadence, the partner base keeps widening, government dollars should arrive in size soon, the balance sheet is in superb condition to weather any near term issues, and the structural question that was live during Q1-namely, can AST make the new bus work in a stacked configuration-appears resolved.

As we wrote in our“Connecting Dots” full thesis, this treasure won't remain hidden forever. Now it's just a matter of execution.

Nebius Group (NBIS)

It's worth pausing to remember where this one sat a year ago. When we first bought NBIS in late 2025, the bear case wrote itself. Nebius was a freshly re-listed carve-out of Yandex, operating a modest data center with a few co-locations across Europe, and a customer book composed almost entirely of VC-backed AI natives and other small, unproven firms. No anchor customer. No enterprise counterparties worth the name. A small but growing fleet of Nvidia GPUs financed with cash the company was burning faster than it was generating. And the elephant in the room was that nobody had any real idea how the capital markets would treat a Russian-adjacent carve-out asking them to underwrite a multi-gigawatt buildout. You had to squint to see a business. What you could see was a team, a collection of good assets arguably trading below liquidation value, and an execution-based timing window.

One year later, the questions that defined that bear case have been answered in sequence, and not one of them broke the wrong way.

Late in 2025, NBIS added META to its customer list with a ~$3B capacity-constrained contract. In March, that became a $27B five-year commitment in two pieces: $12B of dedicated capacity on one of the first large-scale Vera Rubin deployments starting in early 2027, and a further $15B in which Meta commits to backstop Nebius's uncommitted third-party capacity as it comes online. That second piece matters more than the headline suggests, as it turns Meta into a floor buyer for speculative builds and collapses demand risk on capacity Nebius was already planning to scale. Combined with Microsoft, committed contract value now sits at roughly $46B against a platform that did $228M of revenue in Q4. The platform thesis is scaling as we speak, with AAA counterparties.

The sequencing of events also speaks to the company's stellar execution. Five days before the Meta announcement, Nvidia put $2B into Nebius via pre-funded warrants at $94.94 for roughly 8.3% of the company, paired with a partnership targeting 5+ GW of Nvidia systems deployed by end-2030. Nebius has since achieved Nvidia Exemplar Cloud status on GB300 for training, further cementing the preferred-partner relationship. And on cue, the chorus that reflexively cries“Circular capital!” at every AI infrastructure deal cried it again.

In reality, Nvidia is enabling capacity at a preferred customer by paying the reservation fee. Nebius gets priority GPU allocation (still the binding constraint for every neocloud on earth) at zero cash cost and modest dilution. Jensen gets a free option on execution he can self-catalyze. Both sides win, and Nebius walked into the Meta negotiation with silicon supply that its competitors can't match.

Two days after the Meta deal closed, Nebius issued debt in size, and the terms tell you everything about how the bond market views this company. Four billion convertible senior notes upsized from $3.75B on oversubscribed demand, coupons of 1.25% and 2.625%, conversion premiums of 57.5% and 55%. While some other neoclouds dilute their shareholders by multiples of the outstanding count, Nebius is smartly financing growth at low rates and delaying modest dilution. A year ago, we couldn't have told you at what price Nebius could raise debt at all. In Q1, the company printed sub-1.3% paper out to 2031. Last quarter we argued the widening in Oracle CDS had been misread as evidence of AI-capex capital withdrawal; this raise was the confirmation.

Q1 results show a business that, while early in its lifecycle, has clearly already inflected profitably. Revenue hit $399M (up 684% YoY), adjusted EBITDA swung to $130M with the AI cloud segment at a 45% margin, and ARR reached $1.9B. All capacity is sold out, pricing is strengthening, and management reaffirmed $3.0–3.4B in full-year revenue at ~40% EBITDA margins.

On financing, management now expects roughly 60% of 2026 capex to be covered by customer prepayments from Microsoft and Meta, with the remaining 40% from debt and equity. Capex guidance was raised to $20–25B from $16–20B, driven not by cost inflation but by visibility into 2027 capacity that customers have already committed to. Over 90% of the capex range is already secured by cash and contractual commitments. On $16–20B of capex, that's $9.6–12B of customer cash flowing directly into the build and the remaining $6-8B outspend is likely convertible debt, essentially delaying dilution further into the future when the company can more easily absorb it. In sum, the contract-secured structure we described a year ago now has visible terms, and the dilution math on the remaining equity component is far less threatening than the headline capex number would suggest. Cash at quarter-end stood at $9.3B.

Critical to the thesis is power, and here the company showed excellent execution yet again. Contracted power now exceeds 3.5 GW, with a target of at least 4 GW by year-end. The new 1.2 GW owned site in Pennsylvania-Nebius's second gigawatt-scale U.S. campus-will bring 250–300 MW online by end-2027 and full capacity by 2030, joining the Independence, Missouri campus energizing in H2 2026 and a 310 MW site in Lappeenranta, Finland. Owned contracted capacity now represents more than 75% of total power, a deliberate structural advantage in cost and control.

On the software side, three acquisitions this year-Tavily (agentic search), Eigen AI (Nvidia's top-ranked inference speed provider), and Clarifai (system-level optimization)-are being folded into Token Factory, Nebius's inference product. Management frames software as an enabler, as it widens the addressable customer base, improves consumption, and augments margins on the underlying compute. Own the stack, then layer software that makes you the lowest-cost, highest-performance provider.

Nebius's execution has been stellar, and the company is navigating the constrained DC buildout with grace. The pattern demonstrated over the last 12 months (and our expectation going forward) runs like this: secure the physical sites, lock in the silicon, sign the contracts, then layer on the financing to pull the whole platform forward. Each cycle has widened Nebius's advantages versus the field, and we're thrilled to carry it as a scaled position in our book.

FTAI Aviation (FTAI)

A year ago, FTAI Aviation sat in our special situations bucket, navigating a coordinated short-seller campaign that sent the stock into the low-$80s and created an entry that looked uncomfortable in real time, but had the potential to be a long-term holding. Our original thesis was borne out, but also evolved in advantageous ways, so the company graduated from special situation to emerging compounder.

FTAI is a leading MRO franchise for the CFM56 and is transforming into a capital-light, high-visibility model with its Strategic Capital Initiative (SCI), protected by an irreplaceable competitive advantage in PMA parts.

As a reminder, FTAI captures the best economics in aftermarket aviation not by passively leasing engines, but by operating a vertically-integrated platform (Module Factory + SCI) that manufactures“green time” at a structurally lower cost than OEM pathways.“Green time” is simply the remaining usable life on an engine or component before major maintenance; FTAI creates it by tearing down older engines and using proprietary PMA parts and USM components to rebuild them with serviceable modules (fan/core/LPT) that can be swapped in days rather than waiting months for shop visits. The result is both higher aircraft and engine uptime for customers in a supply-constrained market and high-margin Aerospace Products revenue, itself layered on top of leasing, creating an increasingly durable flywheel as scale and parts availability compound.

Full-year 2025 earnings saw Aerospace Products adjusted EBITDA come in at $671M, up 76% versus 2024, and management raised 2026 guidance to $1.625B in total business segment EBITDA. On the supply chain side, FTAI signed a multi-year materials agreement with CFM International, the GE/Safran JV, securing OEM replacement parts, thrust performance upgrades, and component repair capabilities, directly reinforcing its open MRO positioning for the decade ahead. A key point here is that they are partnering with the company from whom they are taking aftermarket share, a sign that the extra-normal profits FTAI generates are not threatening to competitors and suppliers.

FTAI also acquired seven off-lease Airbus narrowbodies from Air France to serve as engine and module feedstock, a now-familiar structure that simultaneously adds inventory and deepens the exchange flywheel. SCI I deployment is now largely complete, with SCI II launched and an anchor commitment already secured.

Q1 2026 confirmed the acceleration: $325.6M in adjusted EBITDA on $830.7M in revenue, with module production nearly doubling year-over-year. Management reaffirmed $1.625B in 2026 guidance, raised the dividend for a third consecutive quarter, and upsized the revolver from $400M to $2.025B, positioning itself to absorb sale-leaseback opportunities as fuel costs and geopolitical stress squeeze airline liquidity and flush engine supply into FTAI's feedstock pipeline.

FTAI Power, the company's effort to repurpose CFM56 engines into aeroderivative gas turbines for data centers, has moved from announcement to execution. Prototype testing is ahead of schedule, a packaging JV with Jereh Group is signed, and management expects 2027 production to be sold out in the near term with a meaningful portion of 2028 already spoken for.

To highlight this new vector of growth again, FTAI is converting a jet engine into stationary power, which does not require reinventing the machine. The process largely involves removing the fan, converting the combustor and fuel system from jet fuel to natural gas, reconfiguring the exhaust section for stationary operation, and packaging the unit to deliver shaft power to a generator. It is not the most efficient solution, but it is often not the most expensive either. When“time-to-power” is the binding constraint, readily available supply matters most. This business line sits incrementally on top of the core MRO franchise and highlights management's willingness to identify adjacent opportunities and move decisively when opportunity knocks.

The core business is executing at scale, a potential threat from GE/Safran looks minimal, guidance is reaffirmed, and optionality is expanding through SCI II and FTAI Power, meaning the position has materially de-risked from our original entry. We remain constructive.

Quarterly Operational Update

We are pleased to share a few operational updates as we continue investing in the firm's foundation. Effective January 1, NAV Consulting formally took over as the Fund's administrator, a transition we believe meaningfully upgrades Crossroads' back-office infrastructure, investor reporting, and overall partner experience. We also recently welcomed Aidan Cyrus as Operations Associate, adding bandwidth across operations, reporting, and day-to-day execution to support a growing partnership.

Finally, the Crossroads team will be in Dallas from Wednesday, May 20 through Friday, May 22 for meetings with existing limited partners and prospective partners interested in learning more about the Fund. If you are in the Dallas area and would like to meet with us while the team is in town, please reach out to George Thornton at [email protected].

Conclusion

Q1 was the kind of quarter that tests whether the work you've done actually holds up when the environment gets uncomfortable. We believe ours did. The portfolio held its ground in a market that offered plenty of reasons to panic, and several of our highest-conviction names continued to move in the right direction while broader indices declined. That said, we are under no illusion that one quarter proves anything. The“suspended animation” we described above will resolve eventually, and when it does, the next developments may not be pleasant. Our job is to own businesses where the outcome depends on company-specific execution and hard catalysts, rather than to guess which way the next headline will break. That's what we've built, and that's what we intend to keep building. As always, we are grateful for your trust and partnership, and we look forward to updating you as the year progresses.

Best regards,

Ryan O'Connor

MENAFN23052026005205015282ID1111158709