Direct Listings, Explained — Why 2026 Broke the Model
Direct listings skip underwriters and bookbuilding entirely. 2026 had more of them than any year on record — and most lost the bulk of their value.
Every company in Zorro's IPO pipeline is taking the same basic path to a public listing: hire underwriters, file an S-1, build a book of institutional demand, set a price, sell new shares at that price. A direct listing skips most of that. There is no underwriting syndicate buying shares to resell, no roadshow, and historically no new shares created at all — existing shareholders simply register their stock and let it trade. Spotify popularized the structure in 2018; Slack followed in 2019, Palantir and Asana both used it in 2020, and Coinbase, Roblox, Squarespace, Amplitude and Warby Parker all went public this way in 2021. For a decade, it was a narrow tool for a specific kind of company: well known, cash-rich, with no real need to raise money or to have a bank drum up retail demand.
2026 changed what the tool gets used for, and the results are a clean case study in why the skipped steps exist.
The "reference price" isn't a price
A traditional IPO sets one number through bookbuilding and sells shares at it. A direct listing has no equivalent. Instead, exchanges publish a reference price — an estimate, usually built off recent private-market trading or the company's last funding round — that isn't a price anyone actually pays. It's a starting point for an opening auction the morning of listing, where a designated market maker matches real buy and sell orders into a single opening print. That print can land far from the reference.
Roblox is the textbook example of this working as intended. The NYSE set a reference price of $45 ahead of its March 10, 2021 direct listing. The stock opened at $64.50 — 43% above the reference — and closed its first session at roughly $69.47, putting the company's fully diluted value near $45 billion. Nothing was "wrong" with that gap; it just reflected genuine public demand finding a price with no bank-negotiated number anchoring it first.
What a direct listing skips, and why that's the point
A traditional IPO bundles in several mechanisms a direct listing doesn't have:
- No underwriting discount. Banks in a bookbuilt IPO buy the shares and resell them for a fee — typically 5–7% of proceeds. A direct listing still uses bank advisors, but nobody is committing capital to buy the stock first, so there's no spread on a trade that never happens.
- No greenshoe. Underwriters in a traditional deal can oversell up to 15% of the offering and buy shares back in the open market to lean against a falling price in the first weeks. A direct listing has no over-allotment to stabilize anything.
- No pre-negotiated institutional anchor book. Bookbuilding exists partly to line up large holders who are comfortable sitting on the stock. A direct listing's buyers on day one are whoever shows up to the open auction.
For a company with a large, liquid float and genuine public name recognition — Spotify, Coinbase, Roblox — none of that machinery was doing much useful work anyway; the stock would have found a fair price regardless. For a small company with a thin float and no institutional sponsorship, removing all three at once is a different proposition entirely.
2026: the record year that also lost the most money
Direct listings had their biggest year ever in 2026 — 13 US companies used the structure, more than any prior year, according to Renaissance Capital. But the cohort's average return from first trade is roughly -71%, and only one of the thirteen is trading above its own opening price.
That one exception, Ionic Digital (IOND), shows what the structure looks like when it still works: shares began trading on Nasdaq on July 28, 2026, opened at $50, and closed the first session at $62.90 — up almost 26% from the open, with real two-sided demand setting the price.
The rest of the cohort looks more like 20/20 BioLabs (AIDX), which completed its Nasdaq direct listing on February 19, 2026 at an $11.42 reference price, briefly traded as high as $50 in the weeks after, and by late August had fallen back to roughly $0.50 a share. That is a company that never built a real institutional shareholder base, never had a bank-negotiated price floor, and never had a greenshoe standing by to absorb selling — on a float thin enough that a handful of early holders cashing out can move the stock by double-digit percentages in a session.
The pattern is the same one SPACs taught
This is, structurally, the same lesson blank-check IPOs taught a few years earlier: a single label — "direct listing," like "IPO" — can describe two cohorts of companies that behave nothing alike, and averaging them together describes neither. Roblox and 20/20 BioLabs both "went public via direct listing" in a literal sense. One event involved a company with a 600-million-user product and years of institutional ownership finding its price in public for the first time. The other involved a thinly held microcap with no underwriter-negotiated price, no stabilization, and no lockup discipline finding out, in real time, how little support a stock has when none of that exists.
None of this makes a direct listing inherently riskier than an IPO — the mechanism is neutral; what matters is what kind of company is using it, and why. A company skipping underwriters because it doesn't need them is a different story from one skipping them because no bank would build a book for it. Zorro's recent-IPO tracker shows first-day returns and lockup timing for every listing in the last three months, traditional and direct alike — the structure a company chose is one more data point worth checking before reading too much into any single day's print.