The reckoning we anticipated last year arrived. Rates rose more aggressively than markets had priced, the crypto industry experienced its deepest contraction yet, and the public markets repriced growth at speed. We are, in some sense, relieved. This is the kind of environment in which our model works best.
On the Year FTX Was Always Going to Happen
The November failure of FTX was, by November, surprising in its specifics and unsurprising in its general shape. A company whose business model rested on the willingness of unsophisticated counterparties to accept opaque accounting was always going to fail; the only question was the timing and the eventual size of the loss. The eventual size was meaningfully larger than the consensus estimate, which is itself a useful data point about how much of the cryptocurrency industry's apparent mass was contingent on this single firm's solvency.
We had no exposure. We continue to believe in the underlying technical claims of decentralized consensus. We continue not to believe in most of the companies that purport to commercialize them. The two positions are not contradictory; the underlying technology has produced — and will continue to produce — durable infrastructure, while the dominant operational businesses in the space have, for several cycles now, replicated the structural problems of the early-1900s wildcat banking era.
The most consequential question raised by FTX, for our work, is not about cryptocurrency. It is about how a generation of sophisticated venture capital firms came to participate in a structure whose audited financials were not meaningfully different from a Ponzi scheme. The answer, we suspect, is the same answer we have given in past letters about other periods of misallocation: the social proof of seeing capable peers participate overrode the diligence that the participating firms would otherwise have applied. The lesson — that diligence must be insulated from peer-firm activity — is not new. It is, perhaps, more important now than it was a year ago.
On the Multiples That Have Returned to Earth
Public-market software multiples, peak to trough this year, contracted by more than sixty percent in the median case. Private-market late-stage rounds have, with a lag of approximately nine months, begun to reflect the same compression. We are now writing checks at terms that, twelve months ago, would have been considered conservative to the point of unfundable.
The compression has not been uniform. Companies with durable unit economics and clear paths to profitability have repriced less than companies with growth metrics and weaker fundamentals. The bifurcation is the kind we have been waiting for; it permits underwriting decisions that price the underlying business rather than the cohort. Several of our 2022 commitments are in companies whose 2021 valuations we declined to participate in; the founders' willingness to accept the lower 2022 prices reflects, in our reading, a recalibration that will compound favorably across the cycle.
On the Founders Who Have Quietly Become More Serious
The founders we are now meeting are, with very few exceptions, more rigorous about their own businesses than the founders we met in 2021. The change is partly composition — many of the less-rigorous founders have stopped raising — and partly behavioral — the surviving founders have absorbed the lessons of the year and incorporated them into their forecasts. Both shifts are durable in our experience.
The behavioral shift is the more interesting of the two. We are now seeing founder cohorts that, eighteen months ago, were operating at headcount levels and burn rates that assumed continued capital availability, who have, in 2022, restructured their operations around eighteen months of runway and growth-at-cost rather than growth-at-any-price. The transitions have been painful; many of these companies have laid off staff. The founders who have made the transitions cleanly are, we believe, the founders we want to back in their next rounds. The founders who have made them poorly — slowly, with denial, with unforced errors — are revealing themselves in ways that compound across our diligence framework.
On the First Checks We Are Now Writing Again
Our pace of new commitments accelerated through the second half of the year. We made eleven new investments in the third and fourth quarters, against four in the first half. The acceleration was not strategic; it was the consequence of prices moving into a range our framework would underwrite. The 2022-2023 vintage will, we believe, be one of our best.
The eleven new commitments are concentrated in three sectors: applied AI (four commitments), enterprise infrastructure (three commitments), and biotechnology with materially shortened development timelines (two commitments). The remaining two are in adjacent categories that we will discuss in 2023 or 2024 letters. The thesis underlying all eleven is the same — that the companies positioned to compound across the next decade are being founded right now, by founders whose first formative experience as operators is the 2022 correction, and whose business plans are calibrated against the conditions our framework was built to underwrite.
A Closing Note
The drought has been long and we have welcomed it. We will welcome it as long as it lasts. The 2023 letter will be an earlier check on whether our 2022 deployment was as well-calibrated as we believe it to have been.
The Partners
Winzheng Family Investment Fund · December 2022