Databento
The Pre-IPO Deep Dive — a $97M NEA-led Series B, revenue growing several-fold a year off a base nobody has seen, profitable at 24 people, and a valuation that has never been disclosed.
Databento is the first credible developer-first attack on the $50B market-data stack: normalized real-time and historical feeds from 70+ venues, full order-book depth, colocated capture, one API, usage-based pricing — sold to engineers at quant funds, prop shops and AI labs rather than to analysts on terminals. The operating facts are unusually strong for a 38-person company: 6.65× revenue growth, 97% enterprise retention, profitability reached before touching the prior round, and a $97.0M Series B led by NEA confirmed on EDGAR against $300M+ of demand. The facts it has not shared are the ones an IPO needs: absolute revenue, valuation, and a CFO. This is an early-stage infrastructure compounder with a 2029+ listing horizon — and the most interesting name in private market data precisely because the public incumbents grow 3–13% a year.
The Numbers at a Glance
1 · Business Overview
An HFT partner's answer to the market-data tax
| Metric | Value |
|---|---|
| Founded | 2019 (as Elisify, Inc.) · Salt Lake City, Utah · public launch 2023 |
| CEO / President | Christina Qi (co-founder; ex-partner, HFT fund Domeyard) |
| Co-founder | Luca Lin (director; ex-Domeyard head of trading) |
| CFO | None · first VP of Finance being hired mid-2026 |
| Board | Qi · Lin · Geoffrey Bernstein · Rick Yang (NEA, joined July 2026) |
| Employees | 24 at profitability · 38 (CNBC, July 2026) · fewer than 30 engineers |
| Disclosed Equity Raised | ~$127–131M (company ~$127M; Form D sum $131.3M incl. conversions) |
| Latest Round | $97.0M Series B · NEA lead · first sale 2026-04-29 · Form D 2026-07-09 |
| Valuation | Undisclosed in every round · no secondary mark exists |
| Revenue | Undisclosed · 985% (2024) → 4.2× (2025) → 6.65× (2026) YoY growth |
| Customers | 3,000+ companies (July 2026) · ~80,000 users (CNBC) · 97% enterprise logo retention |
| Scale | 70+ venues · 20+ PB served · 100+ PB storage secured · 3M+ symbols |
Christina Qi ran a high-frequency fund and spent a career paying the market-data tax: per-venue licenses, per-feed handlers, colocation cross-connects, and months of normalization engineering before a single strategy could be backtested. Databento is that stack sold as a product — one schema, one API, one bill — with exchange license fees passed through "with no upcharge" and margin earned on the infrastructure. Named customers run from Morgan Stanley, DRW, Clear Street and Belvedere to Google, OpenAI, Snowflake and NVIDIA (a customer and a technology partner) — the AI-lab logos are the tell that the addressable buyer has widened from quants to anyone training on tick data.
2 · Product Suite & Moat
Depth, breadth and proximity — the three things an API wrapper cannot fake
| Product | Description | Strategic Role |
|---|---|---|
| Historical data | Usage-based from $0.50/GB; full order book (MBO), trades, OHLCV, statistics; 19 PB normalized | Land motion · self-serve wedge |
| Live data | Subscriptions (CME $199–$4,500/mo; US equities unlimited $4,000/mo; ICE venues $2,500 each); real-time gateways | Recurring revenue · expand motion |
| US Equities & Options | 15 exchanges + 30 ATSs (Jan 2025); all 18 options exchanges via OPRA — "beyond 40 Gbps" processed in real time | Breadth vs Polygon-class peers |
| Global futures | CME group venues; ICE Futures US (Jun 2025), ICE Europe (Dec 2025); Eurex (Nov 2025); EEX (Dec 2025); CFE (Apr 2026) | Exchange-breadth moat · multi-asset |
| PCAPs | Lossless packet capture 1–2 switch hops from the venue handoff; FPGA capture; PTP <25 ns; min order $3,000 | Tier-1 HFT credibility |
| Reference data | Corporate actions + security master for 200+ venues (since Aug 2024) | Stickiness · workflow completeness |
| DBN + clients | Databento Binary Encoding; Python, C++, Rust; QuantConnect LEAN integration (Apr 2026) | Developer lock-in |
| Coming: spot FX | 100+ pairs, 40+ liquidity providers | Next asset class · no fixed income or crypto yet |
The moat thesis rests on four pillars:
- Colocated capture is capital, not code — feed handlers at Equinix NY4, CyrusOne Aurora and Equinix FR2, with LD4, HKEX, JPX and a dozen more sites being onboarded toward "20+ data centers within six months." A pure API reseller buys a consolidated feed; Databento takes the raw handoff. That is what the Series B is paying for.
- Full-depth normalization is the expensive part — one schema across 70+ venues' order-book semantics is years of engineering that customers have already ripped out of their own stacks (an Architect CEO quote lists Reuters, Activ, Bloomberg, Exegy, Redline, QuantHouse and Quincy as the displaced incumbents).
- Usage-based land, subscription expand — $125 of free credits and per-GB history convert students and small funds; usage-based live data was deprecated in 2025 so that live revenue became subscription. 97% enterprise logo retention says the expand motion works.
- Pass-through pricing is a trust moat — publishing exchange fees at cost makes the margin stack transparent, which is exactly what procurement at a quant fund wants to see and what Bloomberg/LSEG cannot offer.
3 · Financials & Unit Economics
Growth multiples without a base — and a Form D that confirms the capital
Databento reports growth, not revenue. Every Form D marks revenue "Decline to Disclose." What it has said: revenue grew 985% year-over-year in the year to October 2024 (with 7,000+ new customers); 4.2× GAAP revenue growth with a record Q3 at the November 2025 strategic round; and 6.65× year-over-year at the July 2026 Series B, with active API users more than doubling "in recent months." It also says it reached profitability "well ahead of schedule" with 24 employees, "without dipping into our last round of funding."
What the numbers do and do not tell you:
- Growth is decelerating in multiple, not in dollars. 10.9× → 4.2× → 6.65× off a base that roughly 45×'d across those three readings means each year's absolute increment is larger than the last. The wobble is in the multiple, as it should be.
- Profitability at 24 people says gross margin is high. With exchange fees passed through at cost, the contribution margin on a $4,500/month CME subscription is overwhelmingly infrastructure-amortized. This is the best unit-economics signal the company has given.
- Pricing power is being exercised — CME Standard moved $179 → $199 in June 2026, explicitly "driven by higher license fees introduced by several exchanges." Whether that is pass-through or margin is the question an S-1 would answer.
- Absolute revenue is the gap. Aggregator estimates in the single-digit millions are inconsistent with a $97M round oversubscribed 3×, NEA leading, and a customer list that includes Morgan Stanley and OpenAI. We do not publish a number because none exists.
4 · IPO Status & Timeline
Series B is the start of the build-out, not the end of the private life
We checked EDGAR directly. Databento Inc. (CIK 0001784241; Delaware; formerly Elisify, Inc.) has five filings, all Form D or D/A — August 2019, March and November 2021, November 2024, and July 9, 2026. There is no S-1 and no public draft registration statement. No outlet has reported bankers, a confidential filing or timing; the CEO's July 2026 CNBC interview did not raise a listing.
A company that still has to hire its first finance leader is not filing
The critical nuance: the Series B is sized for a capex cycle — 20+ data centers, 100+ PB of storage, a layer-1 switching upgrade — not for a pre-IPO balance-sheet dress. NEA's Rick Yang joined the board; the company was simultaneously advertising a "zero-to-one" VP of Finance role with "no team to inherit." Public-company readiness starts after that hire, not before.
- Every round has been unpriced in public; a company planning a listing usually starts anchoring valuation in the press a round or two before.
- The November 2025 strategic round from "a major quant hedge fund" and the Series B's "$300M+ of demand" both say private capital is abundant — there is no liquidity pressure forcing a listing.
- The comparable exits in this niche have been acquisitions, not IPOs: ICE bought Interactive Data for $5.2B, Nasdaq bought Quandl, Exegy rolled up Vela, Enyx and NovaSparks. An exchange or data major buying Databento is at least as likely as an S-1.
Base case: a Series C in 2027–2028 that finally carries a disclosed valuation, a listing no earlier than 2029. Bull case: revenue compounding at multiples makes a 2028 IPO possible. Bear case: an exchange-group acquisition pre-empts the public route entirely.
5 · Competitive Landscape
The incumbents grow at single digits; the API-native peers have no disclosed scale
| Competitor | Ticker | FY2025 Revenue | Market-Data Growth | Market Cap | QL Signal |
|---|---|---|---|---|---|
| Databento | Private | n/d | 6.65× | n/d | — |
| CME Group | CME | $6.52B · data $803M | +13% | $96.0B | Sell |
| Intercontinental Exchange | ICE | ~$6.0B adj. · FI&DS $2.4B | +10% | $88.3B | Neutral |
| Nasdaq | NDAQ | $5.25B · Data & Listing $804M | +7% | $53.9B | Buy |
| MSCI | MSCI | $3.13B | +9.7% | $41.1B | Buy |
| FactSet | FDS | $2.32B (FY Aug-25) | +5.7% organic ASV | $10.6B | Neutral |
| Thomson Reuters | TRI | $7.48B | +3% | $45.5B | Buy |
| LSEG (Data & Analytics) | LSEG.L | n/d here | +5% organic | — | — |
| Massive (ex-Polygon.io) · dxFeed · Exegy | Private | n/d | n/d | n/d | — |
Two competitive truths sit in that chart. First, the incumbents are Databento's suppliers — CME, ICE and Nasdaq sell it the feeds it normalizes, and their 7–13% market-data growth is partly the fee inflation Databento passes through. Second, nobody in the API-native tier has disclosed scale: Massive (the former Polygon.io) has raised single-digit millions; dxFeed sits inside Devexperts; Exegy is a private roll-up of Vela, Enyx and NovaSparks. A $97M NEA round makes Databento the best-capitalised independent in its tier by a wide margin.
6 · The Exchange-Fee Problem
The supplier sets the price of the product
Pass-through pricing is honest — and it means the margin is hostage to CME, ICE and the SIPs
Databento's pricing page says exchange license fees are passed through "with no upcharge." That transparency is a sales asset. It is also a constraint: on June 22, 2026 the company raised its CME Standard plan from $179 to $199 a month and repriced other plans, "driven by higher license fees introduced by several exchanges." Ten days later the SEC approved the CT Plan consolidated-tape fee schedule with a 15.95% inflation adjustment on non-display fees and derived data newly fee-liable — the framework under which Databento had positioned itself as a competing consolidator.
- Upstream leverage: CME's market-data revenue grew 13% in 2025; its own 10-K lists regulatory change to market-data fees as a risk in the other direction. The exchanges will price to the edge of what downstream vendors can absorb.
- Downstream elasticity: Databento's wedge customers — students, small funds, API hobbyists — are the most price-sensitive buyers in the market. Pass-through inflation lands on them first.
- The mitigant is breadth: 70+ venues means no single exchange's fee decision moves the blended price much, and the Series B's global data-center build diversifies the supplier base further.
Why it matters for the IPO: public investors price market-data businesses on gross-margin durability. An S-1 will have to show margin net of pass-through — and show it holding through an exchange-fee cycle. That is the number the company has, and has not shared.
7 · Valuation Framework
Bounding a number the company will not give you
| Metric | Value | Comp / Basis |
|---|---|---|
| Disclosed valuation | None | All five Form Ds silent; no secondary mark |
| Series B size | $97.0M | Form D · includes converted securities |
| Implied post-money (dilution assumption) | ~$0.5–1.0B | $97M ÷ 20% → $485M · ÷ 10% → $970M · QL roster estimate ~$600M |
| Revenue | Undisclosed | Growth 6.65× YoY · profitable |
| Incumbent multiples (market cap / revenue) | — | CME ~14.7× · MSCI ~13.1× · ICE ~14.7× adj. · NDAQ ~10.3× · TRI ~6.1× · FDS ~4.6× |
| Comparable exits | — | ICE–IDC $5.2B (2015) · LSEG–Refinitiv ~$27B (2019) · Nasdaq–Quandl n/d · Exegy–NovaSparks n/d |
Without revenue, multiples are a thought experiment, so we state the assumptions. Late-stage growth rounds of this size typically sell 10–20% of the company, which brackets the post-money at ~$485M to ~$970M. The incumbents trade at 5–15× revenue for single-digit growth; a profitable data-infrastructure business compounding at multiples would command the top of that range or above at IPO — which is to say a $600M–$1B mark is consistent with revenue anywhere from ~$40M to ~$100M. Fair-value triangulation under those explicit assumptions:
- Bull case ($2–3B by 2028): growth holds above 3× for two more years, margin net of pass-through prints above 70%, FX and a second continent of colocation make it the default feed layer for AI-era quant; a Series C prices at unicorn-plus and the 2029 IPO benchmarks to MSCI-style multiples.
- Base case ($0.6–1.0B now, $1.5–2B at the next priced round): growth moderates to 2–3× as the base scales; profitability is maintained through the capex cycle; the next round discloses a number for the first time.
- Bear case (strategic sale at $0.5–1B): exchange-fee inflation compresses the wedge, a large customer concentrates revenue, and CME, ICE or Nasdaq buys the distribution rather than competing with it — a good outcome for the cap table, no IPO.
8 · Key Risks
What public-market diligence will price in
| Risk | Severity | Mitigant |
|---|---|---|
| Exchange license-fee inflation | High | 70+ venue breadth dilutes any one exchange; pass-through is disclosed rather than hidden |
| Disclosure gap (revenue, valuation) | High | None until a priced round or S-1; first finance leader being hired |
| Customer concentration | Medium | Single quant-fund deal "expected to generate millions"; 3,000+ companies and a usage-based tail diversify |
| Fixed-cost step-up | Medium | 20+ data centers and 100 PB storage funded by a $97M round; profitable before the build began |
| Pricing pressure from free / cheap APIs | Medium | Depth, PCAPs and colocation are not replicable by a REST wrapper; 97% enterprise retention |
| Reliability at scale | Medium | Public live-feed complaints in 2026 alongside strong support reviews; status page transparent |
| Key-person | Medium | Qi is CEO, President and signatory on every filing; Lin and an NEA director on the board |
| Litigation / regulatory | Low | None found; CT Plan competing-consolidator rules are an opportunity as much as a risk |
9 · Pre-IPO Exposure Routes Today
No secondaries, no fund proxies — own the suppliers or wait
| Route | How | Caveats |
|---|---|---|
| Private secondary platforms | Listing pages exist on a few venues | No observed trades, no marks, transferability unknown — effectively no market |
| CME / ICE / NDAQ (the suppliers) | Own the exchanges whose market-data revenue Databento's growth partly flows to | Category beta only; QL engine reads CME Sell, ICE Neutral, NDAQ Buy |
| MSCI / FDS / TRI (the multiples) | The listed data-vendor set a Databento IPO would be priced against | No Databento exposure; 3–10% growth profiles |
| NEA-adjacent vehicles | NEA is the Series B lead | Institutional LP access only · diffuse |
| Patience | Wait for the first priced round, then the filing | Base case: no listing before 2029 |
Bottom Line
Databento is the best private asset in market-data infrastructure and the least IPO-ready company in this series. The operating evidence is genuinely strong — 6.65× growth, 97% retention, profitability at 24 people, a $97M NEA round confirmed on EDGAR and oversubscribed three times, and a customer wall that runs from Morgan Stanley to OpenAI. What is missing is every number a public investor needs: revenue, valuation and a CFO. The Series B funds a global colocation build, not a listing, and the sector's exits have been acquisitions by the very exchanges Databento buys feeds from. For growth portfolios this is a structural watchlist name: the next priced round will be the first real valuation signal, and an S-1 is a 2029 conversation. Until then, the investable expression is the supplier set — and the recognition that the incumbents growing 3–13% a year are about to meet a competitor growing several hundred percent.