Kaiko's $110M Round: The Cap Table Says More Than the Press Release

On a Tuesday in the middle of a cycle that has already priced in every possible institutional adoption story, the most consequential transaction of the week was not a token launch. It was not an L2 airdrop. It was a $110 million equity round in a company that sells API subscriptions.

The anomaly deserves to be stated precisely. Kaiko, a market data provider founded in Paris in 2014, closed a Series B extension at $110 million. S&P Global led. BNP Paribas, Nasdaq Ventures, RBC, Coinbase Ventures, and DRW Venture Capital followed.

No token. No public sale. No governance forum. No points program. A private company that packages order book snapshots just raised more capital than the median DeFi protocol has ever held in its treasury.

I read funding rounds the way I read contracts. Cap table first, product second, and then the detail nobody puts in the press release.

Context

Kaiko's business is unglamorous, and that is the entire point. It ingests tick-level trade and order book data from more than one hundred centralized and decentralized venues, normalizes it, timestamps it, and resells the result as a subscription feed. Funds use it to mark positions. Exchanges use it for surveillance. Index providers use it so that a number on a screen can survive an audit.

The competitive set splits cleanly. CoinGecko and CoinMarketCap own the retail terminal. Amberdata and Coin Metrics fight for the institutional tier. Messari sells research layered on top of data. Kaiko sits in the middle of that bracket, and it owns one asset nobody can buy retroactively: history.

This is what makes data businesses structurally strange. You cannot reconstruct a 2017 order book you did not record in 2017. Depth compounds. Every quarter the vendor keeps operating, its dataset becomes harder to replicate — not because the code is clever, but because time has passed. In my 2020 liquidity work I built a Python scraper against Uniswap V2 and tracked over five hundred pairs. Three weeks to write. Ninety days before the history was deep enough to be useful. That gap is the moat, and it is also why incumbents in this sector rarely get displaced by better software.

Institutional grade is a term retail crypto uses loosely. In data, it has a specific meaning: an uptime SLA measured in nines, sub-second latency across venues, a documented methodology behind every derived field, and — most critically — the ability to replay history exactly as it appeared on the day of the trade. Replay is what separates a terminal from a source of record. My first serious audit, back in 2017, meant walking Zilliqa's genesis contracts by hand until I found an integer overflow in the sharding batch logic. Precision over speed is not a slogan in verification work. It is the deliverable. We filed the patch, and the team delayed mainnet by two weeks. Nobody remembers the delay now. Everybody depends on the fix.

The second structural fact is customer type. Every investor in this round is a data consumer, not a data speculator. S&P Global builds indices. BNP Paribas and RBC run risk systems. Nasdaq operates a listing venue. Coinbase operates an exchange. DRW runs a market-making desk. When your investors are also your users, the round is less a bet than a procurement decision with equity attached.

Core

Here is the problem Kaiko actually solves, and why the S&P Global lead matters more than the headline number.

Institutional portfolios need a reference price. A pension fund marking a $200 million crypto position cannot mark it to a single venue's last trade. That price is manipulable, venue-specific, and — as anyone who watched the 2022 stETH depeg or the 2021 exchange outages knows — briefly fictional. The fund needs a composite: a methodology that weights venues by liquidity, filters wash trades, excludes outlier prints, and outputs a number that a regulator, an auditor, and a counterparty can each reconstruct independently.

Building that composite is not a coding problem. It is a provenance problem. Metadata holds the provenance the price ignored. Every tick needs a source, a timestamp, and a survivorship record — and the methodology needs to be documented well enough that a third party can rerun it two years later and reach the same answer.

I learned the shape of that failure in 2022. When the Luna collapse rolled through the market, the funds that got hurt worst were not the ones with the most leverage. They were the ones marking positions against prices they could not defend. I rebuilt our correlation matrix that quarter to strip out single-venue prints entirely, and the model surfaced the hidden leverage links between Celsius and Three Arrows before either firm admitted insolvency. The lesson stuck. The reference rate is the risk model.

That is the layer S&P Global is buying into. If the index giant intends to publish crypto benchmarks — and its acquisition history says it does — it needs a data spine a compliance officer will sign off on. Kaiko does not compete with CoinGecko. Kaiko is a component inside somebody else's regulated product.

Tracing the ghost liquidity behind the rug pull is a weekend exercise. Auditing a reference rate that a pension fund uses to mark a nine-figure book is a five-year project. These are different industries wearing similar names.

The B2B2B structure also explains the syndicate's odd composition. Nasdaq Ventures and Coinbase Ventures are upstream suppliers in disguise: both operate venues whose raw data flows into Kaiko's pipes. Their equity buys visibility into how their own data gets repriced downstream. For S&P Global, the stake buys a seat at the table where the standard gets written. For BNP Paribas and RBC, it buys early access to a vendor their compliance departments will eventually require anyway.

Following the exit liquidity to its cold storage is usually a euphemism for a founding team cashing out. Here it reads literally in a different sense: the capital went into a company that stores history, and history is the one asset that does not exit.

There is one more layer worth naming. In 2026 I led the integration of machine learning models into our fund's anomaly detection stack, trained on five years of on-chain data. The system flagged a $50 million synthetic volume scheme across a cluster of new L2 networks — fabricated prints, coordinated timing, single-entity wallets behind multiple venue identities. We reported it. The transparency frameworks that followed were built on the assumption that third-party verification exists.

That is the whole thesis, and it is unflattering to everyone: the reason institutional money now flows to independent data vendors is that venues have repeatedly misreported their own numbers. A vendor's independence is the product. Coinbase cannot sell Coinbase data to S&P Global as a neutral benchmark. That is precisely why Coinbase Ventures invests in someone else doing it.

Contrarian

Now the part the announcement leaves out.

Funding is not revenue. A $110 million Series B extension tells you what sophisticated allocators believe about the next five years. It does not tell you what Kaiko billed last quarter. Backing out a plausible multiple for a data business at this stage — three to five times forward revenue — implies an annual run rate in the low tens of millions. Real, but not enormous. Anyone reading this round as confirmation that crypto data is a hypergrowth sector is reading a sentiment indicator and calling it a financial statement.

Correlation is not causation, and endorsement is not adoption. Coinbase Ventures writing a check and Coinbase incorporating Kaiko's feed into institutional products are two separate events. Only the second shows up in revenue. Watch for that, not for the announcement.

The structural risk is upstream capture. Exchanges sit on the raw data. Kaiko aggregates it. Aggregators have historically been squeezed the moment their suppliers decide the margin is worth internalizing. Binance already runs internal market data products. Coinbase has the infrastructure to do the same. If either decides to sell normalized feeds directly to institutional clients, the middle layer compresses overnight. Coinbase Ventures on the cap table is an endorsement and an option. Those are not mutually exclusive.

There is a second-order version of this risk that gets less attention. If the round's strategic investors converge on a single vendor's methodology, the industry inherits a standard no competitor can dislodge on merit alone. Data businesses drift toward winner-take-all dynamics — not because one product is better, but because standards are sticky and switching costs are measured in compliance reviews rather than dollars.

Notice also who is absent. No a16z. No Paradigm. No top-tier crypto-native financial VC. The cap table is strategic capital: banks, indexers, exchange CVCs, one prop desk. Strategic investors buy different things than financial investors buy. They buy integration, access, and information rights. The practical consequence is that Kaiko's likely exit is acquisition or a conventional listing — not a token. That is a legitimate outcome, and it is a fundamentally different one from what crypto-native allocators underwrite.

One caveat on structure: this was a B-round extension, not a C-round. Extensions usually mean existing investors exercised pro rata rights and the company chose to manage valuation growth rather than reset a headline number. Read it as discipline, or read it as a ceiling. Both are consistent with the disclosed facts.

Takeaway

The signal to watch this quarter is not the funding total. It is whether S&P Global names a crypto benchmark product that cites Kaiko as its data source. If that happens, the reference-rate layer gets defined by a traditional index provider, and every ETF tracking that benchmark inherits the methodology. That is a quiet, durable repricing of who controls crypto's definition of price.

Three things to monitor over the next ninety days. First, any Coinbase or Binance announcement of an institutional data product — that is the moment the middle layer's margin gets tested. Second, competitor raises from the Amberdata and Coin Metrics tier; capital clusters in a sector within two quarters of a marquee round. Third, 13F filings, where data subscription expenditure occasionally surfaces through affiliated vehicles.

The last question rarely gets asked at the announcement. When the price of an asset is determined by a methodology a committee controls, whose methodology is it — and who audits the committee?