Business

Know the Business — Onyx Biotec Limited

Onyx is a single-site fixed-cost factory that converts glass, plastic, API powder and distilled water into ampoules and vials for ~100 Indian pharma brand-owners — paid per unit filled, not per drug sold. The whole P&L is gated by one variable: how full the two Solan lines are running. The market currently prices this at roughly book value (~₹58 Cr market cap vs ₹55 Cr equity, P/B 1.05×). The decision-relevant question is not whether earnings will recover from the FY26 loss, but whether ROCE can stay above the cost of capital for long enough to justify any premium to replacement value.

1. How This Business Actually Works

Onyx charges a per-unit conversion fee (₹ per ampoule or vial filled) and books the raw-material pass-through as revenue. A ₹0.10 swing in conversion fee per SWFI ampoule on 230 million units a year is ₹2.3 Cr of operating profit — roughly two-thirds of FY26's entire operating profit. That is why this is a volume game pretending to be a pharma business. The two units run different economics: Unit I (sterile water, FFS ampoules, capacity 638,889/day) is the high-volume commodity engine; Unit II (cephalosporin dry-powder injections and dry syrups, capacity 40,000 + 26,667/day) is the higher-realisation product attempting to ramp post WHO-GMP certification in May 2024.

The bargaining-power asymmetry is the single biggest fact in this business. The top-10 customers that paid 71% of Onyx's FY24 revenue (top-5 = 51%) are themselves multi-thousand-crore pharma companies — Sun Pharma alone is ~1,000× larger by revenue. They can switch ampoule vendors in 60-90 days; Onyx cannot switch them. The Solan SME competitor 30 km down the road offers the same ampoule at a few paise less and runs the same WHO-GMP file. The two ways out are (a) scale to Akums-level breadth across 30+ dosage forms, or (b) acquire EU-GMP/USFDA status that opens regulated export pools. Onyx has done neither.

About 25-35% of the cost base is fixed once the cleanroom and FFS line are commissioned. That is the operating-leverage equation: a 10-point swing in utilisation moves OPM by roughly 4-6 percentage points. FY26 is the live demonstration — revenue grew 12% to ₹69.5 Cr but operating profit collapsed two-thirds because employee cost, depreciation and other opex stepped up with Unit II ramp while utilisation lagged.

2. The Playing Field

The peer set shows two clusters and one outlier. Pure-play scale CDMOs (Innova, Senores) earn 15% ROCE at 15-27% OPM. Branded-plus-CMO hybrids (JB Chem, Cohance) trade at premium multiples because the branded layer earns 25-30% margins regardless of plant utilisation. Akums sits in the middle — scale CDMO at lower margin but real ROCE. Onyx is the outlier on every dimension that matters: 1.1% ROCE is roughly one-thirteenth of the peer median, and the P/B at 1.05x already reflects this — the company trades at less than half the price-to-book of Akums and around an eighth of JB Chem's. The lesson is not that Onyx is cheap; it is that the market has correctly priced an unproven sub-scale CDMO at scrap-plus-working-capital value while peers are priced on demonstrated capital efficiency.

The competitive landscape is even harsher than this table suggests. Onyx's RHP names its top customers as Hetero, Mankind, Sun Pharma, Aristo, Macleods, Akums itself, and Reliance Life Sciences — every one of these companies either runs its own injectable lines or buys from larger CDMOs at better prices. The closest economic substitute is not the listed peer set but the ~1,000 unlisted Solan/Baddi WHO-GMP SMEs that compete at the SWFI ampoule end. Schedule M (effective 31 December 2025 for sub-₹250 Cr units) should thin that tail and is the single meaningful structural tailwind for compliant CDMOs like Onyx — but only if the displaced volume is large enough to push Onyx's utilisation past 75%, which has not yet happened.

3. Is This Business Cyclical?

This is not a macro-cyclical business in the way a steel or chemical company is. The cycle is utilisation-driven and customer-specific. The two recent regime changes prove the point. FY22→FY23 saw revenue fall 12% (₹44.9 Cr → ₹39.5 Cr) and OPM compress only 150 bps — that was the post-COVID end of vaccine-dilutant SWFI demand and the textbook example of a customer-order reset in this layer. FY26 looks different but works the same way: revenue rose 12% on Unit II ramp, but operating profit fell 67% (₹10.4 Cr → ₹3.4 Cr) because raw-material cost-of-goods absorbed a higher mix shift before pricing caught up, and fixed costs (D&A flat at ₹3.05 Cr, employee cost stepping up) stayed on the books regardless.

The order is the lesson. A reader who tracks only debtor days and Unit I utilisation will know about a margin reset one to two half-years before it hits the bottom line. The H1 FY26 print (net loss ₹0.9 Cr on ₹34.6 Cr revenue) was already the warning that the full-year FY26 result would be a write-off; the consensus narrative of "Unit II ramp = margin expansion" was wrong by the time the September 2025 print arrived.

4. The Metrics That Actually Matter

The three metrics that matter most are capacity utilisation, customer concentration, and debtor days — in that order. Every other number is downstream. EBITDA margin is the consequence of utilisation; net income is the consequence of EBITDA margin minus depreciation; ROCE is the consequence of net income relative to a near-fixed capital base. The mistake an analyst will make on this name is to anchor on revenue growth (the FY26 revenue print of ₹69.5 Cr looked fine) instead of margin and working-capital trajectory (the cash conversion cycle has nearly tripled from 51 days in FY22 to 119 in FY26).

ROCE was approaching peer-respectable levels (12.85% in FY24, 12.16% in FY25) before falling off a cliff in FY26 — a single-year collapse of more than 11 percentage points. A peer like Innova posts 15% ROCE consistently because its cost base is amortised across many more dosage forms and customers. The FY26 print tells you that the prior two years were not the new normal; they were the IPO-window peak before Unit II's depreciation and ramp costs collided with a soft revenue mix.

5. What Is This Business Worth?

The right lens here is price-to-book against trailing ROE, plus a tangible-asset and working-capital floor — not a forward P/E, not a DCF on Unit II ramp, not an EV/EBITDA multiple borrowed from Akums. The reason is that earnings power for a sub-scale single-site CDMO is inherently unstable across a 5-year window, but the replacement cost of the cleanrooms, FFS lines and the ~₹50 Cr of net working capital is real. The market cap of ₹58 Cr against book equity of ₹55.4 Cr (FY26 BS) and fixed assets net of ₹39.2 Cr plus other assets of ₹52.3 Cr is essentially: "pay 1x book for a real asset base earning a negative spread over the cost of capital today, and decide if the option on Unit II reaching 75% utilisation is worth more than zero."

Sum-of-the-parts is not the right lens here — there are no listed subsidiaries, no investment portfolio, no regulated rate-base, no holding-company stack. The two manufacturing units are reported as a single primary segment per Ind AS 108, and even if they were split, Unit II is too early in its ramp (one full year post-WHO-GMP) to be valued as a separate business. The single asset to underwrite is the same single asset to operate: 357 employees in Solan running two production lines for a fragmented Indian pharma customer base.

6. What I'd Tell a Young Analyst

Watch three numbers in every half-yearly disclosure and stop trying to model the rest: Unit I capacity utilisation, debtor days, and the top-5 customer share of revenue. The first tells you whether Schedule M is delivering displaced orders. The second is the leading indicator of customer financial stress — when brand-owners stretch payables, it shows up at their CDMO before it shows up in their own filings. The third tells you whether the diversification trend (top-5 from 80% to 51% over FY22-FY24) is continuing or reversing.

The market is most likely to mis-price this name in two directions. The overestimate: treating the FY24-FY25 ROCE recovery to ~12% as the new run-rate and applying a peer CDMO multiple (P/E 30-45×); the FY26 collapse to 1.14% ROCE has now disproved that. The underestimate: the Schedule M tail-thinning of unlisted Solan SMEs is a slow, two-to-three-year structural tailwind that the SME-tier market typically does not price until utilisation actually moves; if Onyx prints 70% Unit I utilisation in FY27 with stable debtor days, the setup for a re-rating from P/B 1.05× toward the 1.8-2.2× book band would be on the table.

The thesis changes if (a) a top-3 customer leaves and revenue concentration spikes to 80%+ again; (b) the company files for EU-GMP or USFDA at Unit II — a multi-crore capex that opens a new economic engine; (c) promoter holding at 65.10% changes by more than 3 percentage points in any direction (insider exit signal or strategic stake build); or (d) two consecutive half-years of negative CFO drain working-capital cushion below ₹30 Cr, at which point the asset-floor argument starts breaking. Until one of those happens, this is a book-value name with optionality, not a CDMO compounder.