FTAI Infrastructure is a Fortress-managed infrastructure conglomerate that underwent radical transformation in 2025: two acquisitions ($1.05B Wheeling & Lake Erie Railway, full consolidation of Long Ridge Energy & Power) funded with $2.25B of new debt and preferred equity. The result is a company with $5.75B in assets, $3.8B in debt, negative total equity of $(146M), and a 10-K that explicitly states current liquidity is insufficient to meet the $218M Jefferson bond maturity due July 2026.

The headline financials look attractive until you decompose them. Revenue grew 29% to $503M. Adjusted EBITDA tripled to $361M. But $233M of that EBITDA — 64% — comes from Long Ridge, a 485MW gas-fired power plant in Ohio that management is "actively evaluating for strategic alternatives, including a potential sale." Strip Long Ridge out and you're left with $128M of EBITDA servicing $3.8B of debt. Interest expense alone was $266M in 2025 and is rising — the $1.35B Term Loan refinanced in February 2026 carries a 9.75% coupon.

The capital structure is a layer cake of claims senior to common equity: $3.8B debt, $1B Ares preferred with 1.5x mandatory redemption ($1.5B by 2029), $160M GCM convertible preferred, and Fortress extracting management fees regardless of performance. Common equity sits at the absolute bottom of a $5.5B claim stack.

The company knows this is tight. The Backstop Agreement for the Jefferson bonds was signed on March 16, 2026 — the same day the 10-K was filed. That's a deliberate legal maneuver to neutralize the going concern disclosure. The underlying liquidity problem is deferred, not solved.

The Bull Case: Long Ridge Is the Entire Company

The thesis is simple: Long Ridge is worth more than FIP's market cap, and management is trying to sell it.

Three gas power plant transactions in the last 18 months set the comps. Constellation acquired Calpine at 7.9x EBITDA ($985K/MW). NRG bought LS Power's gas fleet at 7.5x ($923K/MW). These are large, diversified portfolios — single-asset premiums typically run higher.

At 7.5-8x on Long Ridge's $233M EBITDA, the implied enterprise value is $1.75-1.86B. After stripping out approximately $1B of Long Ridge-level debt (Senior Notes, credit facility, CanAm loan), the equity value flowing to FIP is $750-860M. FIP's market cap is roughly $700M. Long Ridge alone, at infrastructure multiples, covers the entire common equity value — before accounting for the remaining business.

There is genuine optionality beyond the base comp. Long Ridge sits on approximately 1,660 acres on the Ohio River in PJM — the most supply-constrained power market in America. PJM capacity auction prices surged roughly 10x from $29 to $329/MW-day, driven by data center demand and retiring baseload generation. The 10-K mentions "artificial intelligence data centers" as a potential use for the site — one sentence, no contracts, pure optionality. But Talen sold a similar nuclear campus (960MW, 1,200 acres) to AWS for $650M plus a 1,920MW PPA through 2042. Long Ridge has the same ingredients: power, land, water, grid interconnection, and PJM location.

The CEO appears to agree with the bull case. Kenneth Nicholson bought 500,000 shares at $5.22 in May 2025 — $2.6M of personal money, open market. The CFO bought twice. These are people who see the Long Ridge sale process books and the full capital structure. Insider buying during active "strategic alternatives" is not a gesture.

The Bear Case: The Waterfall Eats Everything

The bear case isn't that Long Ridge is worthless. It's that Long Ridge proceeds never reach common equity.

Start with the claim stack. A Long Ridge sale at $1.8B EV generates roughly $800M of equity after retiring Long Ridge's own debt. That $800M needs to address: the $1.35B Term Loan (9.75%, due Feb 2028), Ares' $1.5B preferred redemption (2029), remaining Jefferson and Repauno debt, and Fortress's management fees. The math doesn't close cleanly. Partial de-leveraging helps, but the remaining capital structure still overwhelms the remaining EBITDA ($128M ex-Long Ridge).

The Fortress governance pattern is not theoretical. New Fortress Energy, founded by Fortress co-founder Wes Edens, is in severe distress as of early 2026: missed interest payments on $2.7B of secured notes, trading at 28 cents, creditor groups formed with Evercore and Perella Weinberg advising. FTAI Aviation — FIP's sister vehicle — paid approximately $300M to terminate its Fortress management agreement in May 2024 and subsequently re-rated higher. Muddy Waters shorted FTAI in January 2025 alleging misleading accounting, triggering a 24% drop and audit committee investigation.

The pattern across Fortress-managed vehicles is consistent: complex capital structures, aggressive leverage, external fee extraction, eventual reckoning. FIP's management fee runs approximately 13% of market cap annually with no performance hurdle. Every dollar of preferred equity issued to solve the liquidity problem increases Fortress's fee base.

Operating cash flow was negative $118M in 2025. Not adjusted EBITDA — actual GAAP operating cash flow. The company is financing-dependent. The Backstop Agreement buys 364 days on the Jefferson maturity, but the terms are undisclosed and the underlying cash generation doesn't cover the obligations.

And then there's the railroad. Transtar's revenue declined $7.1M year-over-year on lower carloads, even before the Wheeling acquisition filled in the headline number. The 15-year Railway Services Agreement with USS has passed its minimum commitment period — there are no volume guarantees from year 6 onward. The Ares preferred sits at the railroad subsidiary level, meaning Ares has a $1.5B senior claim on Transtar and Wheeling assets before FIP corporate sees a dollar.

The Correction: Nippon Steel Changes the Railroad Picture

One material fact the 10-K understates: Nippon Steel completed its acquisition of USS in June 2025 under a Trump Executive Order, with a National Security Agreement that includes $11B of investment commitments by 2028 and government consent rights on plant closures. Gary Works Blast Furnace #14 is getting a $350M reline that extends its operating life by 20 years. Granite City is restarting an idled blast furnace.

This structurally reduces the Transtar volume risk that the bear case relies on. USS plants won't close — the government has consent rights. Production is being sustained and expanded, not rationalized. The electric arc furnace transition threat is real long-term but pushed out by decades of blast furnace investment.

Short-term headwind: the Gary BF reline requires a 100-day outage (May-August 2026), which will temporarily reduce Transtar carloads and look ugly in the Q2 10-Q. This could create a panic entry point for those who understand the context.

Where the Edge Is (and Isn't)

Factor decomposition reveals FIP's structural problem for sizing. Measured idiosyncratic variance is 65.5% — below the 75% threshold. Market beta is 1.99. Every dollar of FIP carries two dollars of market exposure.

The edge-driven variance is approximately 57%: Long Ridge sale catalyst (≈30%), Fortress governance pattern recognition (≈8%), insider signal (≈5%), and partial edge in power market / railroad analysis (≈14%). The remaining 43% is factor noise — market beta, credit spreads, energy prices — where we have no informational advantage.

The regression understates the true idiosyncratic content because the dominant driver is a latent catalyst. Long Ridge sale isn't in the historical returns. Pre-catalyst, FIP trades on beta and credit conditions. Post-catalyst, it's a completely different company. The factor decomposition tells you what FIP has been, not what it becomes.

But you pay beta rent while waiting. In a 15% market correction, FIP drops approximately 30% from beta alone — before any idiosyncratic deterioration. The bear case target of $2 becomes plausible from factor exposure alone.

What the Market Implies vs What We Think

Reverse-engineering the stock price against our scenario framework:

The market gives FIP a 44% chance of reaching $2 (near-wipeout), a 40% chance of muddling through to $7, and a 16% chance of the bull case ($14). We think it's 30/40/30. The 14-percentage-point swing from bear to bull is where the alpha lives.

The options market confirms this skew. January 2027 puts outnumber calls 3.7:1. $3 strike puts had 150 contracts traded on filing day. July 2026 put implied volatility runs 24% above call IV — an unusual and severe downside skew concentrated around the Jefferson bond maturity. The market is most afraid of the near-term liquidity test and more balanced on the 18+ month horizon where Long Ridge could resolve the picture.

The probability-weighted EV is $7.60 from $5.62 — a 35% raw return, roughly 17.5% idiosyncratic alpha after adjusting for edge percentage. But the distribution is violently bimodal. This is not a stock where expected value tells the full story. The 10th percentile path matters more than the mean.

Conclusion

FIP is a catalyst play wrapped in a distress wrapper. The alpha is not in the 10-K — the filing is net bearish and the market has priced the distress signals correctly. The alpha is in the cross-ticker corroboration: Long Ridge is worth $1.75-1.86B at current gas power M&A multiples, PJM capacity prices have structurally shifted in its favor, and the CEO bought $2.6M of stock during the active sale process.

The market prices a 16% probability of the bull case. We think it's 30%. That 14-percentage-point gap, concentrated in a single binary catalyst, is the entire position.

The right expression is small and defined-risk: 1.5% max weight, options preferred over stock (no beta rent, defined downside), entry at $5.00-5.25 (at the CEO's anchor), with July 2026 Jefferson maturity as the first resolution point and December 2026 as the time limit.

If the Long Ridge sale materializes at comp multiples, the 19% short interest and 15 days to cover create conditions for a violent squeeze. If it doesn't, the capital structure grinds common equity toward zero and you lose your premium. This is a bet you take 20 times at the right price. The question is whether you can get the entry right.

Five open questions remain before full conviction: Long Ridge Senior Notes change-of-control provisions (do sale proceeds get trapped at the subsidiary level?), Jefferson Backstop terms (how expensive is the bridge?), Ares noncompliance triggers (when can Ares seize the railroad?), actual buyer pipeline for Long Ridge, and Gary BF outage impact on Q2/Q3 railroad earnings. These are HIGH priority gaps. The thesis is interesting enough to track and position for at small size, but not enough to size with conviction until the waterfall math is clear.

Evidence