Hudson Technologies (HDSN): the refrigerant reclaimer the screen sees as a melting ice cube

Hudson Technologies earned $0.37 a share in 2025. Three years earlier it earned $2.20. A trailing-earnings screen divides today's price by the most recent number, gets a P/E near 17, and if it bothers to pull up the trend at all sees five years of net income moving from $32M to $104M to $52M to $24M to $17M — the shape of a melting ice cube. Screens don't have a column for "commodity price spike followed by normalization followed by regulatory floor."

Coverage Gap

Hudson Technologies

is classified by financial data providers as "Specialty Chemicals" — the same category that catches diversified chemical manufacturers, adhesives companies, and catalysts businesses. (The mislabeling problem is not unique to HDSN; Innospec (IOSP) runs the same categorization risk with three structurally different end markets under the same label.) That label is wrong in a way that matters: Hudson is a refrigerant services and reclamation company. It collects used refrigerant from industrial and commercial customers, purifies it to AHRI purity standards in its own certified laboratory, and resells it. No new chemical synthesis; no manufacturing plant; no patent-protected molecule with a cliff. The model is closer to a specialty waste-to-resource business than to a chemical company.

The mislabel accelerates the screen's error. After the 2022 price spike — pandemic-era refrigerant shortages drove net income briefly above $100M — prices normalized and earnings collapsed. A momentum or quality screen that sees a "specialty chemicals" company with four consecutive years of declining net income exits without examining whether the underlying business and its regulatory context have changed. Between the wrong sector tag and the unflattering earnings trend, Hudson carries almost no sell-side coverage

Business

Hudson's operations run through one reportable segment. The core activity is buying back used refrigerant — from commercial HVAC contractors, industrial processors, refrigerant wholesalers, and the United States military — purifying it, and selling it back as certified reclaimed refrigerant

The company sells both reclaimed refrigerant and some virgin (newly manufactured) refrigerant, alongside what it calls "RefrigerantSide® Services" — on-site system decontamination, diagnostic testing, and energy-efficiency work. Revenue in FY2025 was $246.6M, up 4% from $237.1M in 2024 on higher volumes, partially offset by lower average selling prices.

The U.S. Defense Logistics Agency (DLA) is a key customer. Hudson has held a prime contractor role for military refrigerant supply since 2016. The most recent five-year contract — awarded October 2025 — is currently subject to a bid protest filed by a competitor, and the DLA has rescinded the award pending its internal review. The existing contract runs through July 2026; Hudson is continuing logistics support in the interim

In 2024, Hudson acquired USA Refrigerants, a refrigerant distributor, and in December 2025 added Refrigerants Inc. ($2.2M), a Denver-based distributor and reclaimers. The 10-K describes both as expanding Hudson's access to recovered refrigerant for its reclaim pipeline.

Moat / Mispricing

Refrigerant reclamation is not a moat business in the conventional sense — Hudson competes with "numerous regional and national companies" and wins primarily on equipment performance, service breadth, and price

What it may have, and what the screen cannot see, is a regulatory tailwind that structurally differentiates reclaimed refrigerant from virgin supply.

The AIM Act directs the EPA to phase down U.S. production and consumption of hydrofluorocarbon (HFC) refrigerants. The initial step-down was 10% from baseline levels in 2022 and 2023. The Act then mandates a cumulative 40% reduction in the baseline from 2024 through 2029

Crucially, "there are no restrictions placed on the reclamation of HFC refrigerants" — reclaimed refrigerant is explicitly outside the allowance system.

In September 2024, the EPA went further: the Refrigerant Management Rule, implementing Section (h) of the AIM Act, includes a mandate requiring "the use of reclaimed HFCs for certain applications." This is the structural demand-floor mechanism: as virgin production falls under the allowance caps, certain end-users are required by regulation to source reclaimed refrigerant rather than simply choosing between it and virgin supply. Hudson, which the 10-K describes as "one of the largest refrigerant reclaimers in the United States," is a primary supplier of that mandated product.

The mispricing hypothesis follows: the 2022 spike was a COVID-era dislocation (supply shortages drove spot prices and margins far above sustainable levels). The 2023-2025 decline is the normalization — the right response by any screen. But the screen cannot distinguish "normalization to pre-spike baseline" from "normalization to a regulatory-supported new floor" because both look like the same falling earnings trend. If the Refrigerant Management Rule's reclaim mandate begins to tighten the effective supply of compliant refrigerant for covered applications, the floor is real and the trailing earnings are the wrong starting point for valuation.

Capital Allocation

The balance sheet is clean. At December 31, 2025, the company held $39.5M of cash and equivalents against negligible net debt — net interest income for 2025 was $2.5M, not expense, because the balance sheet is essentially unlevered

Management's capital allocation in 2025 sends a readable signal. The company repurchased 2,890,240 shares for $20.0M — an average of roughly $6.92 per share, close to the current price and well below book value per share (equity of $243.4M on roughly 42M shares implies book of approximately $5.79). It also built inventory from $96.2M to $135.9M, an increase of $39.7M, funded by the drawdown in operating cash flow

The two moves together — buybacks at trough and a large refrigerant inventory build at year-end — are consistent with management positioning for a 2026 price recovery in reclaimed refrigerant. They could also be wrong.

Risks

Self-checking the desk's usual blind spots:

• DLA contract risk: The U.S. military contract is both a concentration risk and a near-term known unknown. The existing contract runs to July 2026; the newly awarded replacement is on hold pending bid-protest review. If the DLA's internal review results in a rebid that Hudson loses — or an extended limbo — a meaningful customer is disrupted at exactly the wrong moment for the thesis. The 10-K discloses the protest without quantifying the revenue share, which is itself a flag for investors to research independently.
• Margin compression with no visible floor: Gross margin went 38.6% (2023) → 27.7% (2024) → 25.2% (2025), driven by lower average selling prices and higher freight costs. The 2025 report attributes the fall to lower prices for "certain refrigerants" but does not name them. If the specific refrigerants losing price are in the HFC categories that the Refrigerant Management Rule will tighten, that's a thesis argument. If they are HFO or legacy HCFC refrigerants not covered by the mandate, it's a separate headwind.
• Inventory risk: $135.9M of refrigerant held at year-end is stated at lower of cost or net realizable value (FIFO). A continued price decline would require write-downs through cost of sales. The 10-K explicitly lists commodity price risk as a principal market risk, and the company has experienced inventory impairments in past cycles. A write-down year would push reported NI below the already thin $16.7M 2025 level.
• Negative 2025 OCF: Operating cash flow turned to -$3.2M in 2025 from $91.8M in 2024 — almost entirely the $39.7M inventory build. If the thesis is right, that cash comes back as the inventory sells at recovered prices in 2026. If the thesis is wrong, the company funded a bad bet with cash that could have stayed on the balance sheet.

Valuation

At $6.23 and roughly 42M shares, the market cap is approximately $262M. On trough 2025 net income of $16.7M, that is about 15.7x trailing earnings — not cheap for a business with falling margins and a major customer contract in limbo, but not the reflexively expensive read the screen implies from the 17x P/E on $0.37 EPS. On equity of $243.4M the stock trades at roughly 1.08x book, which is historically low for a company this cash-generative in a normal year (2024 OCF was $91.8M).

The valuation case depends entirely on what earnings recover to. If gross margins stabilize near 27-28% (the 2024 level, before the 2025 SG&A spike that included $4M of severance) and revenue holds near $250M, operating income normalizes to something like $25-30M and net income to $18-22M. That is an earnings recovery of 10-30% from 2025 — modest, and it produces a stock that looks roughly fairly valued at the current price. The upside scenario — where the reclaim mandate translates into genuine pricing power and gross margins step back toward 35-38% — requires both regulatory enforcement and continued HFC phase-down, and is not guaranteed. The 2022 peak was a dislocation, not a reference rate; the honest central case is that reclaim becomes structurally better than 2025 trough but does not revisit 2022.

Re-rating Catalyst

The screen re-rates the stock when trailing earnings stop looking like a trend and start looking like a data point. That requires a visible gross margin recovery — probably to 27-28%+ for at least two consecutive quarters — which would tell the screen that 25.2% was the floor, not a direction. The H1 2026 10-Q (expected August 2026) is the first data point for this test. Resolution of the DLA bid-protest removes the largest known uncertainty; any EPA enforcement guidance on mandatory reclaimed-HFC applications would crystallize the demand-floor argument. Until one of those is visible in reported numbers, this stays a bet on regulatory tailwinds that a trailing-earnings model cannot price — which is the coverage gap. The cyclical-trough P/E pattern at work here is one of three cases documented in Three cases where the trailing P/E ratio actively misleads.