Read this stock off the consolidated line and you mis-price it in both directions.

• Four businesses moving at four speeds. India core services ($338M NTV, +4.1% Adj EBITDA / NTV, +26% YoY in Q4) and International ($75M NTV, just breakeven) are profitable engines. Native durables ($37M NTV, −8.9%) and InstaHelp domestic-help ($8.4M NTV, −210%, $4.76 loss per order) are deliberate burns. The consolidated EBITDA line is the arithmetic average.
• IndiaMART multiple on a Swiggy operating profile. UC trades at 12.1× P/S vs IndiaMART 7.7× and Naukri 19× (profitable) — and Swiggy 3.1×, Eternal 4.3× (burning). Same revenue base as IndiaMART ($166M vs $167M); −16% OPM vs IndiaMART's +30%. The premium prices the destination, not the print.
• Core + International + cash already cover ~75% of the cap. India core at 3× NTV (~$1.01B) + International at 3× NTV (~$224M) + $350M net cash and treasury ≈ $1.59B against a $1.97B market cap. The remaining ~$380M is what the market is paying for Native + InstaHelp optionality.

The market sold the InstaHelp loss line. The same release printed the bear's named refutation.

• What the market priced. Stock −11% on May 8 after a Q4 PAT loss of $17M versus consensus profit of $1.6–2.3M; InstaHelp loss-per-order widened from $4.06 to $4.76 even as orders grew ~1.7× sequentially (1.6M → 2.7M) — the optical 'moat does not transfer to housekeeping' data point.
• What it ignored. India core NTV grew +26% YoY in Q4 — the fastest in 11 quarters — accelerating from 19% (Q2) to 21% (Q3) to 26% (Q4), straight through Goldman's 24% terminal CAGR ceiling. The bear's explicit stop-loss is two quarters of sub-15% core growth. Q4 did the opposite, on supply-constrained hours-per-partner.
• The cleanest moat proof in the listed set is uncited. The FY18 cohort still spends 1.93× its year-1 base seven years later; the FY20 cohort survived COVID at 1.04×. None of the three published foreign initiations — Ambit Sell $1.01, Morgan Stanley UW $1.22, Goldman Neutral $1.46 — references cohort retention. The consensus debate is about InstaHelp; the moat decision is in the core.

A subsidy war UC must win on either economics or attrition. So far the economics are getting worse.

• The arms race. Snabbit is approaching ~1M monthly bookings (per Moneycontrol / Morgan Stanley note) on a $56M April 2026 raise — closing in on UC's InstaHelp 1.1M March 2026 print. Pronto hit ~500k monthly orders (18K/day in March 2026) and raised $20M in May 2026, doubling valuation to $200M. UC InstaHelp loss-per-order widened $4.06 → $4.76 at ~1.7× sequential order growth (1.6M Q3 → 2.7M Q4) — the literal opposite of network economics.
• UC's edge is the war chest. $350M deployable ($215M cash + $135M treasury) versus Snabbit's $49M cumulative raise and Pronto's $18M — 7× and 20× the rival balance sheets. Treasury yield alone (~$14M/year) covers 64% of FY26's $23M free-cash burn. Endurance moat, not unit-economics moat.
• Q2 FY27 (~Nov 2026) is the verdict print. Bull thesis requires loss-per-order below $3.66 with core NTV holding 22%+. Bear thesis requires above $5.23 plus a Snabbit or Pronto round at ≥$700M valuation. Management thinned forward disclosure on this exact metric in Q4 — segment EBITDA ÷ order count is the proxy now.

The FY25 'first profitable year' the IPO marketed was 88% deferred-tax credit. FY26 reverted.

Strip the $25M deferred-tax credit and $14M of treasury yield out of FY25 and the operating business lost $5M — the DRHP itself flagged the tax line as non-recurring. FY26 reverted to a $25M loss and an $11M operating-cash outflow as InstaHelp burn accelerated. The accounting is clean — no auditor change, no working-capital tricks, ESOP excluded from Adj EBITDA is the only metric-hygiene flag. What does recur is the India-core Adj EBITDA / NTV walk: −22.5% (FY22) → +4.1% (FY26) on hours-per-partner rising from 59 to 90 per month.