Everybody says the supply chain needs data. Nobody is buying it. They're buying the companies.

On 29 November 2022, Maersk and IBM published the obituary of the most ambitious data platform the shipping industry ever built. TradeLens had more than a hundred participants, two of the most credible names in the business behind it, and a mandate that sounded unarguable: make global trade visible. Read what its own founders wrote when they switched it off, verbatim:

"TradeLens was founded on the bold vision to make a leap in global supply chain digitization as an open and neutral industry platform. Unfortunately, while we successfully developed a viable platform, the need for full global industry collaboration has not been achieved. As a result, TradeLens has not reached the level of commercial viability necessary to continue work and meet the financial expectations as an independent business."

Note what that sentence does not say. It does not say the technology failed. It says the industry never showed up.

Two and a half years later, in April 2025, a different kind of transaction closed in the same industry: $241 million in cash for a maritime data business — not a platform, not a consortium, a company. And in June 2026, more than a billion dollars went into the buyer.

Same data. Same industry. One model dead, the other priced like infrastructure. Edition #5 is about why — and the answer is uncomfortable for us.

1. What the machine heard

A reminder of how our engine works, because the method is the argument. We call it signal-first. We never start from an opinion about where the valuable data is. We start from facts — dated, sourced, external. A port investment. A freight capacity move. A regulator opening a file. Every signal is a real article from the trade press, with a URL and a date, or it does not exist for us.

Between 28 May and 2 August 2026, our map collected 1,877 dated press signals across 60 publications, resolving to 532 data holders. Sorted by trade-press family, freight, logistics and ports come back with 198 signals — 10.5% of everything we collected — tied to 122 holders.

Now the part that a marketing team would cut, and that we are printing instead.

Those 198 signals come from 14 publications, and 74% of them come from two: FreightWaves (48%) and Supply Chain Dive (26%). That is not the sound of an economy. That is the sound of an editorial beat. We know exactly what that concentration is worth, because we checked the other candidates for this edition and killed three of them on the same test: robotics came in at 66% from a single outlet, tariffs at 62% from a single outlet. A number that big, sourced that narrowly, is a chamber with an echo in it — and a cluster we cannot tell apart from one newsroom's assignment desk is not a market signal. So we did not build this edition on the volume. We built it on what the money did, which is a different and much harder thing to fake.

2. The obvious story, and why it died in public

The pitch writes itself, and half the industry has heard some version of it: the supply chain is famously opaque; visibility is a universally agreed problem; therefore there is a market for supply-chain data; therefore build the neutral platform where it gets exchanged.

That story has already been tested, at Maersk's and IBM's expense, with a hundred participants and a decade of goodwill. It died. And it died in the most specific way possible: not from a technical failure, but because the parties who generate the data would not come to the table to share it.

There is a temptation to file this under "too early." Resist it — and here's the honest caveat you should apply to everything above: that obituary was written by the deceased. It is a corporate statement, and it takes care to absolve the technology and blame the ecosystem. We are citing it because it is primary and unambiguous about the commercial outcome, not because Maersk is a neutral witness to why its own venture failed.

But the commercial outcome is not in dispute. And the reason it matters is what happened next, in the same data, with a different shape.

3. The same data, priced like an industrial asset

While the neutral platform was being switched off, the granular maritime data business — vessel identity, ownership, port calls, movements — was changing hands at prices that have nothing to do with a failed market.

Spire Global sold its entire maritime business to Kpler. From Spire's own announcement, verbatim:

"The $241 million transaction consists of a $233.5 million purchase price and $7.5 million for services over a twelve-month period, post close."

The completion date is confirmed by an independent primary source — the UK competition regulator's decision records that "The Merger was completed on 25 April 2025."

Then, on 3 June 2026, Kpler announced "a minority strategic growth equity investment of over $1 billion from Sixth Street." A minority stake. Over a billion dollars.

A word on what we are deliberately not telling you. Several widely-repeated figures put a valuation on that round. We checked the primary announcement: it contains no valuation at all. The numbers in circulation are secondary, attributed to unnamed sources, and they disagree with each other by more than a billion dollars. So we are publishing the one number that exists in a primary document — "over $1 billion for a minority stake" — and nothing else. The same applies to Kpler's revenue and to every multiple derived from it: not in a primary source, not in this edition.

4. The tell: what the seller refused to sell

The most revealing sentence in this entire file is not a price. It is a carve-out.

Spire sold all of its maritime business — and kept one slice of the customer list. Verbatim, from the same announcement:

"Spire will retain its satellite network, technology and infrastructure and will continue to serve its aviation, weather and space services customers, along with the existing U.S. government portion of its maritime customer portfolio."

Read that again with a commercial eye. A seller exiting a business drew a line through its own customer base and kept the government side. What a seller refuses to include tells you what it believes is worth keeping — and it says something precise about who the paying customer for maritime data actually is.

The client lists agree. Windward — another maritime data company, taken private in 2025 — describes its customers verbatim as ranging from:

"energy supermajors, shipowners, mining companies, freight forwarders, and port authorities, to banks, insurers, and governmental organizations."

And on 9 April 2026, Windward acquired a US defense software firm explicitly, in the announcement's own title, "to deepen U.S. defense and national security capabilities."

Notice who is not driving that list. Carriers, terminals and shippers appear on it — we are not going to pretend otherwise, and any edition that told you the logistics industry doesn't buy software would be lying to you. Logistics buys plenty: Descartes Systems, the audited public comparable in this space, reported $729.0 million in revenue for FY2026, up 12%. The industry spends, and its spending is growing.

The distinction is finer, and it is the one that matters: the operators buy tools to run their own business. The premium is paid by the people who need to know things about ships they do not own — traders, insurers, banks, and governments. That is the demand that survives contact with a primary source.

5. The clock — and exactly how much weight it can carry

There is a date on the horizon, and we are going to state it with its caveat attached rather than sell you urgency.

On 10 November 2025, the US Trade Representative suspended its Section 301 port-fee action on China-linked vessels. Verbatim:

"The action will be suspended for one year, and as of 12:01 a.m. Eastern Standard Time on November 10, 2025."

Those fees are assessed against a ship's identity — where it was built, who owns and operates it, its net tonnage — at each port call. That is the mechanism worth understanding regardless of what happens next: it converts a provenance record into an invoice. When a data point becomes the basis of a payment, someone starts paying to be certain of it.

The caveat, stated plainly: a one-year suspension announced in November 2025 runs out in November 2026, and nothing in the primary document says the fees will return. We have no verified basis for predicting that they will. Anyone telling you this is a countdown to a certainty is selling something. It is a reason to have your provenance data in order, not a forecast.

6. The regulator opened the door, then closed it on a technicality

Here is the fact we found last, and it is the one we would build the next twelve months of analysis on.

Exactly one competition authority has looked at consolidation in maritime data. In decision ME/2238/25, the UK's Competition and Markets Authority explains why it opened the file, verbatim:

"The CMA's mergers intelligence function identified the Merger as warranting an investigation, having found that there was a reasonable chance that […] it has resulted, or may be expected to result, in a substantial lessening of competition."

And here is how it ended. The parties revised their own UK turnover figure — the decision notes, verbatim, that they "submitted new information regarding Kpler's UK turnover following the CMA's decision to call in the Merger, including at an advanced stage of its Phase 1 investigation" — and the regulator recorded that "Kpler's published accounts do not provide a geographic breakdown of Kpler's turnover." The conclusion was jurisdictional:

"The CMA does not believe that it is or may be the case that a relevant merger situation has been created because the safe harbour threshold in section 23(2)(c) of the Act is not exceeded."

Strictly literally: the regulator suspected a substantial lessening of competition, the turnover threshold was not met, and the substance was never examined. Not cleared on the merits — never reached. That is not a scandal, and we are not going to inflate it into one. It is a gap in the public record, on the only question that would tell us whether this data is consolidating into something a buyer can no longer negotiate with.

7. The bottom line — and the part that costs us something to write

Put the pieces together and the shape is unambiguous.

The neutral, open exchange for supply-chain data was built by two of the largest companies in the world, and it failed for lack of participants. The same underlying data, assembled and owned by an integrated company, sells for hundreds of millions and attracts billions. The buyers are largely outside the operating industry. And the one regulator who glanced at the resulting concentration closed the file without examining it.

On this axis, data is not catalogued. It is acquired. It changes hands as companies and exclusive feeds, bought by parties who need certainty about assets they don't own — not as datasets on a shelf.

We run a data marketplace. So let us finish the thought properly, because you can check it: for this particular filière, a marketplace is the wrong shape. Freight and maritime data is a market of acquirers. The value sits in an integrated book of record that someone is willing to buy whole and stand behind — not in a listing.

That is a real verdict against our own model on a real cluster in our own map, and we would rather publish it than be caught quietly avoiding it. It is also exactly why the editions where we do claim a marketplace works — operational data from the physical economy, held by operators who have no other route to a buyer — should be worth more to you. A method that never returns an inconvenient answer isn't a method. It's a brochure.

What we do next: we said in Edition #4 that our own labels hid the biggest cluster on the map, and that we trust the outside signal over the inside label. Same discipline here. Three other candidate axes were investigated for this edition and three were killed — including one we expected to win — because we could not name a buyer who pays for the thing the thesis described. If you hold operational data and want to know which of those two worlds you are in, that is precisely the question our engine exists to answer.

→ If you operate physical assets and want to know whether your data has a buyer — or whether it doesn't — put it in front of the engine.

A note on method

Our internal figures come from one source: the signals our own engine collected between 28 May and 2 August 2026 — 1,877 dated press signals across 60 publications, resolving to 532 data holders. The freight family (198 signals, 122 holders) was isolated by keyword classification of press headlines, an exogenous cut: it uses the outside world's words, not our own labels. We do this deliberately, because we learned that several of our internal fields are derived from a single upstream variable and therefore cannot contradict each other — a number that cannot be wrong is not evidence. The 74% two-publication concentration in that family is reported here for the same reason.

Every external claim in this edition was checked against a primary document, and quotes are reproduced only where we read them in the retrieved text. Four candidate axes were researched and independently attacked by a skeptic whose job was to refute them; of 48 assertions judged, 24 were refuted — including several load-bearing claims of the thesis you have just read, which is why some obvious figures are missing above. Where a primary source does not state a number, this edition does not print one.