Okapi Solar — PE/VC Investment Memorandum & Valuation
Conclusion
RISK-GATED PROCEED — proceed to diligence only if 2 of 3 items clear
- Instrument
- $1.0M SAFE @ $8.5M cap
- Defensible EV
- ~US$15–16M
- Diluted return
- ~1.9× / ~16% XIRR
- Outcome band
- ~16–29% XIRR
- Hold period
- 5 years
- Verdict
- Risk-gated proceed
An Investment Committee memorandum on a USD 1.0M SAFE into Okapi Solar, a Malaysian rooftop-solar financier. Its central argument is that the valuation debate is a distraction — the decision turns on the credit book and the funding line, not on whether the company is worth USD 16M or USD 28M. Stripped of the solar-growth narrative, Okapi is a specialty-finance consumer lender: a ~5% net-interest-margin credit fund whose 28–68% ROE is leverage, not skill. The recommendation is a risk-gated proceed — a reasonable early-stage bet if, and only if, two of three diligence items clear.
An independent model and reconciliation
Three methods, a 2× disagreement
DCF and VC converge near US$15–16M; NSV front-loads a decade of cohort value
Independent rebuild of the deal model; DCF shown at the midpoint of its US$15–16M range.
The work builds an independent model rather than accepting the deal memo, and immediately finds the memo and the delivered model disagree: every headline memo number traces to a single stale input — 2030 revenue of MYR 47.9M that the model later cut to 43.3M. The three valuation methods disagree by roughly 2× (VC near USD 14M, DCF at 15–16M, NSV up at 28M), with the memo leading with the richest. The honest, defensible enterprise value is the ~USD 15–16M where DCF and VC converge; NSV only looks larger because it front-loads a decade of cohort value on day one.
Return, not valuation, drives the call
The 6.1× headline return is undiluted; carrying Series B/C/D dilution that a post-money SAFE cannot shield against, and weighting in a 25% failure case, the investor's real expected outcome is ~1.9× / ~16% blended XIRR — regardless of whether the cap is fair.
The deal is also fundamentally a Malaysia investment: the data-centre demand story is real but nearly decision-irrelevant (~+9% kicker, ~90% domestic), and every overseas market is hostile to a rooftop-lease model — Vietnam killed its feed-in tariff, Indonesia abolished net metering, Thailand allows net-billing only, India subsidises ownership.
The three risks that decide it
First, credit: the model assumes 0.0% loss every year on a bureau-excluded, never-stressed book with no equity buffer, where a plausible ~3.4% loss rate erases pre-tax profit. Second, funding: debt at 100% of gross receivables through one facility — an extraordinary, revocable advance rate on unseasoned EM consumer paper, with unhedged EUR/MYR exposure and a 30% balloon just past the forecast window. Third, SAFE terms: the diluted base is ~1.9× / 16%, with the bear intrinsic barely clearing the cap.
These are not independent — an EM stress event fires all three at once through a 100%-levered, FX-mismatched, single-facility book.
What would change the answer
The XIRR outcome band
The faithful base case fails the 25% hurdle; each single relaxation moves it — only an exit re-rating clears it
Scenario relaxations on the diluted base case (~1.9× MOIC).
At a faithful base case the deal returns ~16% and fails a 25% hurdle, but it clears under any single reasonable relaxation: 30% recovery-in-failure lifts it to ~17.3%, a 15% failure probability to ~19.2%, and an exit re-rating to ~29.2% — a realistic outcome band of ~16–29% set by defensible beliefs, not by the valuation. The decision reduces to three pieces of evidence: cohort payment/roll-rate data, the funding-facility term sheet, and the diluted waterfall against the actual cap table.