A Cold-Plate Company for $4.75 Billion — and the Bond Ladder That Paid for It

On March 20, 2026, a company that describes itself as "a global leader in water, hygiene and infection prevention solutions" — $16 billion in annual sales, 48,000 staff, customers in more than 170 countries — agreed to pay $4.75 billion in cash for a Calgary firm that makes cold plates.

Ecolab's press release is a clear document and it does not hide any of this. But a press release is not the only thing a company files when it borrows five billion dollars, and the other filings say things the release has no reason to.

What was bought

CoolIT Systems is, in Ecolab's own words, "a pure-play data center liquid cooling company" — coolant distribution units, cold plates, direct-to-chip loops, rack manifolds. It has been at it more than 25 years, out of R&D centres in Calgary and Taipei, with manufacturing in Canada, China and Vietnam. Ecolab expects it to generate approximately $550 million in sales over the next 12 months, and says the acquisition doubles its Global High-Tech addressable market from $5 billion to $10 billion. The seller was funds managed by KKR.

The strategic logic is not mysterious, and Ecolab states it plainly: chips are moving from air cooling to liquid cooling, liquid cooling means fluid, and fluid is what Ecolab has sold for a century. Chief executive Christophe Beck frames it as covering "semiconductor fabs that manufacture chips, power plants that fuel the chips, and data centers that utilize the chips."

That is a real thesis. This post is not about whether it is right. It is about what the financing documents record.

The ladder

To pay for it, Ecolab went to the bond market in May 2026 and issued $5.0 billion of senior notes in four tranches, all maturing on June 15 of their respective years.

Net proceeds were approximately $4.95 billion after underwriting discounts and expenses. Ecolab had also arranged a delayed-draw term loan in April as a backstop; once the notes priced, it terminated that facility without ever drawing on it.

Which $3.6 billion was actually for the deal

Here is the detail that is absent from the acquisition announcement, because it lives in the terms of the notes rather than in the story about the deal.

Three of the four tranches — the 2029, 2031 and 2033 notes, $3.6 billion of the $5.0 billion — are defined in the prospectus as the "SMR notes". If the acquisition had failed to close, Ecolab was contractually required to redeem them at 101% of principal plus accrued interest. The fourth tranche, $1.4 billion due 2036, carries no such condition. The prospectus is explicit about the failed-deal scenario: "the 2036 notes will remain outstanding."

This matters because the stated use of proceeds is the usual elastic formula — to fund the acquisition "and for general corporate purposes". That phrase alone gives you no way to say how much of the $5.0 billion was really for CoolIT. The special mandatory redemption does. It is Ecolab's own contractual line between money raised for this deal and money raised alongside it, and it puts $3.6 billion on the near side.

Every comparison below uses the $3.6 billion, not the $5.0 billion. It is the conservative choice and it is the company's own.

The arithmetic the multiples imply

Ecolab priced the deal at "29x and 24x estimated next 12-month and 2027 adjusted EBITDA". A multiple and a price together imply the denominator. At 29x, $4.75 billion implies next-12-month adjusted EBITDA of about $163.8 million; at 24x, 2027 adjusted EBITDA of about $197.9 million.

Set those against what the borrowing costs. The three deal-contingent tranches carry $175.65 million of coupon a year. All four carry $250.55 million.

On the narrow, conservative attribution — deal-contingent notes only — the annual coupon of $175.7 million exceeds the implied next-12-month adjusted EBITDA of $163.8 million, by roughly $12 million. Against implied 2027 EBITDA of $197.9 million it falls back under, by about $22 million.

So the crossover happens somewhere in 2027, on Ecolab's own numbers. Which is exactly what Ecolab told investors: excluding non-cash amortisation, the deal is "expected to be accretive to adjusted diluted earnings per share in 2028, with the contribution building significantly thereafter."

The filings and the arithmetic agree. The point of doing the arithmetic is that "accretive in 2028" is a forecast you either accept or don't, while "the coupon is bigger than the EBITDA for the first year" is a subtraction you can check.

The clock started before the asset did

The notes priced in May. The deal closed July 2, 2026. Ecolab was paying interest on the borrowing before it owned the thing the borrowing bought, and the 10-Q puts a number on that gap: a $6.6 million special charge in interest expense in Q2 2026, "related to debt used to fund the CoolIT Systems acquisition."

Meanwhile the company sat on $5.1 billion of cash at June 30 — "we increased our cash balances during the period to fund the CoolIT Systems acquisition" — of which $554 million was held outside the US.

What nobody can tell you yet

The obvious next question is what $4.75 billion actually bought in accounting terms: how much is plant and equipment, how much is developed technology and customer relationships, how much is goodwill. That number would tell you how much of the price was assets and how much was expectation.

It does not exist yet. Ecolab's Q2 10-Q says so directly: "Based on the timing of the close of the transaction, it is impractical to include a preliminary purchase price allocation." The same paragraph declines to present pro forma figures, on the grounds that they are "not material to the Company's consolidated financial statements."

A $4.75 billion acquisition, deemed immaterial for pro forma purposes. That is a statement about Ecolab's size rather than CoolIT's: against $16 billion of annual sales, CoolIT's expected $550 million is about 3.4%. The first allocation should land in the Q3 10-Q, and it is the single most informative document still outstanding on this deal.

What I am not claiming

I have not established Ecolab overpaid. The coupon-versus-EBITDA comparison is a first-year cash-cost observation, not a valuation. It ignores CoolIT's growth — which is the entire thesis, and which Ecolab expects to accelerate Global Water's organic sales growth by 2%. A business growing fast enough makes a first-year negative spread irrelevant.

The implied EBITDA figures are mine, not Ecolab's. They invert stated multiples, and they assume the $4.75 billion cash price equals the enterprise value those multiples were struck on. EBITDA multiples are normally quoted on a cash-free, debt-free enterprise value, which need not equal headline cash consideration. If the two differ, both implied figures move proportionally. Ecolab has not published CoolIT's EBITDA, so this cannot be checked against a stated number. One consistency check does pass: $163.8 million against $550 million of sales is a ~29.8% EBITDA margin, which squares with Ecolab calling CoolIT "high-margin".

Coupon is not interest expense. It ignores the small gap between coupons and effective rates, amortised issuance costs, and the return on that $5.1 billion cash pile. It is the gross cost of the borrowing, which is the thing I wanted to size.

The takeaway

The AI build-out is usually described in megawatts and gigawatts. This is what it looks like as a balance sheet: a century-old water and hygiene company issuing debt out to 2036 to buy the firm that makes the plates that sit on the chips.

The interesting number is not the $4.75 billion. It is the $3.6 billion — the portion of the borrowing Ecolab was willing to make contingent on this one deal. Two things are worth keeping separate there. The 1% premium on a special mandatory redemption is what the deal risk cost: bondholders would have been handed back their principal plus a point had it collapsed. The $3.6 billion is not a price at all; it is an attribution, and it is the clearest statement in any of these documents about how much of this borrowing was really about cooling.

How I compiled this

Every figure above comes from a document Ecolab filed with the SEC: the March 20 8-K exhibit, the May 20 prospectus supplement, and the Q2 10-Q filed August 6. No figure comes from secondary coverage of the deal.

The compiled series is in data/ecolab-coolit/, split into deal.csv and notes.csv. Four rows are arithmetic of mine rather than figures Ecolab published, and they are prefixed DERIVED_ in the file itself so the distinction survives being opened in a spreadsheet by someone who has not read this post. The two charts use different colours for the same reason: reading a coupon off a prospectus and inverting a multiple are not the same kind of fact, and a chart that draws them identically is claiming they are.

One small pleasure of working from filings: the prospectus names the acquisition target as "Frigeo Holdings LLC". CoolIT Systems is the brand. The entity that changed hands has a different name, and only the legal document says so.