SLB Agrees to Buy German Cooling Maker Kelvion for $3.4 Billion in Cash Plus $0.7 Billion in Debt
The Houston-based oilfield-services company said on August 31 it will buy Kelvion from Apollo-managed funds and Triton-advised funds, in a deal it values at about 11 times Kelvion's estimated 2026 earnings before interest, taxes, depreciation and amortization.
SLB Bets $4.1 Billion That Oil Engineers Know How to Cool an AI Data Center
Two press releases went out on August 31, 2026, describing the exact same deal. SLB, the Houston oilfield-services giant, said it was paying $3.4 billion in cash for Kelvion, a German cooling-equipment maker[1]. Apollo, the private-equity firm selling Kelvion, said the deal was worth $4.1 billion[3].
Neither number is wrong. SLB's $3.4 billion is the cash it's writing a check for. Add the roughly $0.7 billion of Kelvion debt that SLB is also taking on, and you get Apollo's $4.1 billion — the value of the whole business, debt included[1][3]. SLB picked the smaller number because it's the buyer, spending real money. Apollo picked the bigger number because it's the seller, and a bigger exit looks better to the investors who fund its next deal[3]. Both are telling the truth about the same transaction, from opposite sides of the table.
That splitframing is a preview of the bigger question sitting underneath this deal: whether SLB is buying its way into the future, or paying a peak price for a boom that could cool off before the ink dries.
A Century-Old Furnace Parts Maker Meets the AI Boom
Kelvion is based in Herne, Germany, and has been making heat-exchange equipment for more than 100 years[1][12]. Its business is moving heat out of places that get too hot — industrial plants, refineries, and now, increasingly, data centers.
That last part is why SLB wants it. The newest AI computer chips run so hot that just blowing air over them doesn't work anymore[10]. Data centers are switching to liquid cooling instead, which pumps fluid past the chips to carry heat away. That fluid still has to dump its heat somewhere, and that's the job of the heat exchangers Kelvion builds[10].
SLB expects Kelvion to bring in $2.3 billion to $2.4 billion in revenue in 2026. Of that, $1.2 billion to $1.3 billion comes from data centers — slightly more than half[1]. The rest comes from ordinary industrial customers, so Kelvion isn't a pure bet on AI. It's a real, profitable manufacturer that happens to sit in the middle of a hot trend.
On profit, Kelvion is expected to earn $350 million to $400 million in adjusted EBITDA — a standard measure of operating profit — on that $2.3 billion to $2.4 billion of revenue[1]. That works out to roughly $1 of every $6 in sales staying on as profit before overhead items like interest and taxes, which is a normal margin for industrial equipment, not the fatter margin software companies post.
SLB is paying about 11 times that 2026 profit figure[1][2]. In plain terms, at today's earnings, the purchase price would take about 11 years to earn back. SLB says $120 million a year in expected savings and cross-selling — combining purchasing, cutting overlap, and pushing Kelvion's products through SLB's sales relationships — brings the effective price down to about 8.5 times[1]. That savings figure isn't just marketing. Once a company tells investors to expect $120 million a year, it's committed to finding it somewhere, which is one reason questions about Kelvion's German workforce matter even though nothing has been announced[1].
Why an Oil-Services Company Wants Out of Oil's Mood Swings
SLB used to be called Schlumberger, and its core business has always tracked the price of oil. When crude is expensive, drillers spend more and SLB does well. When it drops, so does SLB's business. Investors know this, and they price oil-services stocks accordingly — lower than steadier businesses, because the swings are baked in[5].
That's the structural pressure driving this deal, and it's separate from anything SLB says about AI specifically. Every dollar of revenue SLB can shift into a market that grows independent of the oil cycle is a dollar investors may be willing to pay more for[5][8]. SLB has been moving this direction before: it already completed the purchase of ChampionX as part of a broader push beyond traditional oilfield work[5].
The data-center business is growing fast enough to make the case. SLB's existing Data Center Solutions unit grew 63% year over year in the first half of 2026[8]. The company projects the combined data-center business, Kelvion included, could hit $4.5 billion to $5 billion in revenue by 2028, with $700 million to $800 million of adjusted EBITDA[6]. If that holds, SLB will have converted itself, in part, from an oil-cycle company into an infrastructure company — a re-rating that could lift how the market values the whole business[5][8].
The Seller Is Betting the Opposite Way
Here's the part that's easy to miss if you only read SLB's side: Apollo and Triton, the funds selling Kelvion, are financial owners whose entire business model is to buy, improve, and exit[3]. They're choosing to sell cooling exposure right now, at what looks like the strongest bid the market has offered in years.
That's not a knock on either party. It's worth sitting with, though, because SLB and its sellers are taking opposite positions on the same forecast. SLB is buying because it believes AI-driven demand for cooling keeps growing. Apollo and Triton are selling because, from their seat, now is the best price they're going to get[3]. Both bets can be individually rational. They can't both be right about where the AI infrastructure cycle actually stands in 2027 and beyond.
That tension is the real story here, more than either company's press release lets on.
The Number Skeptics Keep Pointing To
The skeptical case isn't that AI is fake. It's about timing and concentration. Capital spending by the 14 largest public data-center operators is running close to $750 billion in 2026[9]. That's an enormous sum committed by a small number of buyers, and three of the four biggest hyperscale companies lost stock market value after recent earnings calls specifically because investors worried the spending was getting too large[9].
If that spending slows down, before or shortly after this deal closes, SLB would be left holding a cyclical industrial asset it bought at a full price. The deal isn't expected to close until the first half of 2027, which leaves more than a year for the AI spending environment to shift before SLB actually owns Kelvion[1]. The sellers, by contrast, get their cash on closing and are done. Whatever happens to AI capital spending after that becomes SLB's problem, not Apollo's or Triton's[3].
SLB's shares did hit a new 12-month high around the announcement, a sign the market liked the deal, at least in the near term[14][15]. Whether that holds depends entirely on which side of the $750 billion question turns out to be right.
What Nobody's Saying Yet
One gap runs through nearly every version of this story, on both sides of the Atlantic: nobody has said what happens to Kelvion's German workforce. The company has already passed through private-equity ownership once, in Herne and other German sites, and is now headed to an American buyer[12][3]. Neither SLB nor Kelvion has published any plant-level, headcount, or works-council commitments. No German or EU regulator has been named as reviewing the deal, even though SLB says the transaction still needs unspecified regulatory approvals before it can close[1].
Coverage of the deal split along predictable lines. Bloomberg and other financial outlets led with SLB's smaller $3.4 billion figure and called Kelvion a "data center cooling firm," adopting the buyer's framing[4]. Trade publications serving the data-center and energy industries more often used Apollo's larger $4.1 billion total and repeated SLB's own growth projections without pushback[6][7]. One outlet, Tech Times, went further, declaring in its headline that "air cooling fails AI math" and oilfield engineers "win" — language that treats a deal still awaiting regulatory approval, more than a year from closing, as an already-settled outcome[10].
It isn't settled. The agreement is signed. The price is public. What it will have bought — a smart pivot into steady growth, or a full price paid at the top of a cycle — is a question the market won't be able to answer until sometime after the first half of 2027, when the deal is set to close[1].
Summary
SLB, the Houston oilfield-services company once called Schlumberger, said on August 31, 2026 that it has agreed to buy Kelvion, a German maker of cooling and heat-exchange equipment[1]. SLB will pay about $3.4 billion in cash and take on about $0.7 billion of Kelvion's debt[1]. Add those together and the whole company is valued at roughly $4.1 billion — which is the number the seller, Apollo, used in its own announcement the same day[3]. Both numbers are correct. They just describe different things, and which one an outlet chose is the first thing worth noticing about this story.
Kelvion is based in Herne, Germany, and is more than 100 years old[1][12]. It makes the hardware that moves heat out of buildings and machines. AI computer chips run hot, and the newest ones run hot enough that blowing air over them no longer works well — so data centers are shifting to liquid cooling, which needs exactly the kind of heat exchangers Kelvion builds[10]. SLB says Kelvion should bring in $2.3 billion to $2.4 billion of revenue in 2026, of which $1.2 billion to $1.3 billion comes from data centers[1]. That is the fastest-growing part of the business, and it is what SLB is really buying.
The genuine dispute is not about the facts of the deal. It is about the bet underneath it. SLB and its supporters argue the company is buying its way into steady, growing demand and away from the boom-and-bust of drilling; its own data-center unit grew 63% in the first half of 2026[8]. Skeptics argue that AI construction spending is itself the thing in question. Capital spending by the 14 largest public data-center operators is running near $750 billion in 2026, and three of the four biggest hyperscale buyers lost market value after recent earnings calls when investors balked at those commitments[9]. If that spending slows, SLB will have paid a full price at the top. The deal is not expected to close until the first half of 2027, and it still needs regulatory approvals that neither company has named[1].
The Event
On August 31, 2026, SLB Limited announced a definitive agreement to acquire 100% of Kelvion Holding GmbH, a thermal management and heat-exchange manufacturer headquartered in Herne, Germany[1]. SLB will pay approximately $3.4 billion in cash and assume approximately $0.7 billion of Kelvion debt, for a total transaction value of about $4.1 billion[1][3]. The sellers are Apollo-managed funds, which hold the majority stake, and funds advised by Triton, which hold a minority stake; SLB is buying both[1]. SLB filed the announcement with the U.S. Securities and Exchange Commission on a Form 8-K the same day, and said the deal is expected to close in the first half of 2027, subject to regulatory approvals and customary closing conditions[13][1].
Undisputed Facts
- SLB announced the agreement on August 31, 2026 and disclosed it to the SEC on Form 8-K[1][13].
- The cash price is approximately $3.4 billion; SLB will also assume approximately $0.7 billion of debt, for a total value of about $4.1 billion[1][3].
- Kelvion is headquartered in Herne, Germany, and was founded more than 100 years ago[1][12].
- Kelvion was majority-owned by funds managed by Apollo Global Management, with a minority stake held by funds advised by Triton[1][3].
- SLB says Kelvion is expected to generate $2.3 billion to $2.4 billion in revenue and $350 million to $400 million in adjusted EBITDA in 2026, with data centers accounting for $1.2 billion to $1.3 billion of that revenue[1].
- SLB values the deal at about 11 times estimated 2026 EBITDA before synergies, and about 8.5 times including expected annual run-rate synergies[1][2].
- SLB projects about $120 million in annual EBITDA synergies within three years of closing, and says the deal will add to earnings per share and free cash flow per share in the first 12 months after closing[1][2].
- The transaction is expected to close in the first half of 2027 and requires regulatory approvals that have not been publicly specified[1].
- SLB previously completed the acquisition of ChampionX as part of a broader move beyond traditional oilfield work[5].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Escaping the oil multiple
- Oilfield-services companies trade at low valuations because their revenue rises and falls with the crude price, and investors discount that swing. Any revenue SLB can move into a steadier, faster-growing market may be valued more highly by the market. This structural pull explains the ChampionX deal and this one, independent of anything SLB says about AI[5][8].
- The heat problem is physical, not narrative
- Densely packed AI chips produce more heat per rack than air can carry away at acceptable cost, which forces a shift to liquid cooling and to the heat exchangers that make it work[10]. That constraint holds regardless of whether AI companies make money, as long as the buildings get built.
- A sponsor exit is a timing signal
- Apollo and Triton are financial owners whose business model is to sell. They are exiting cooling exposure precisely when a strategic buyer will pay the most for it. SLB is buying that same exposure. Both can be rational; they cannot both be right about where the cycle is[3].
- The savings promise creates its own pressure
- SLB has told investors it expects $120 million of annual EBITDA synergies within three years[1]. 'Synergies' means the combined company earns more or spends less than the two did apart — usually through overlapping overhead, shared purchasing, and cross-selling. That public target becomes a commitment management must meet, which is why workforce and plant questions in Germany matter even though nothing has been announced.
Material realityTwo figures describe the same deal, and both are accurate: $3.4 billion is the cash SLB pays, and about $4.1 billion is what the whole business is valued at once the $0.7 billion of assumed debt is counted[1][3]. Kelvion is a real, profitable, century-old manufacturer expected to earn $350 million to $400 million in adjusted EBITDA on $2.3 billion to $2.4 billion of revenue in 2026 — meaning roughly $1 in $6 of sales is left after operating costs, which is normal for industrial equipment, not a software margin[1]. Slightly more than half of its revenue comes from data centers; the rest comes from ordinary industrial and energy customers, so the business is not a pure AI bet[1]. The price of about 11 times 2026 EBITDA is an industrial-equipment multiple: in plain terms, at today's earnings the purchase price would take about 11 years to earn back. SLB says savings and cross-selling bring that effective figure to about 8.5[1]. Nothing has closed. The agreement is signed, the deal needs unnamed regulatory approvals, and closing is expected in the first half of 2027 — which means the AI spending environment can change substantially before SLB owns the asset[1].
Narrative as a weaponThree parties are actively shaping how this reads. SLB wants you to believe it is an infrastructure company that happens to have oilfield roots, so it leads with the smaller cash number, the 63% growth in its data-center unit, and the 2028 revenue projection[1][8][6]. Apollo wants you to believe it built and sold a $4.1 billion global leader, so it leads with the larger total value[3]. The trade press, on both the energy and the data-center side, largely amplifies whichever of those two framings fits its readership, because deal announcements arrive as press releases and the numbers in them are not contested. What almost no one is supplying is the third view: that a private-equity seller and a strategic buyer are taking opposite positions on the same 2027 forecast, and that the wider record — roughly $750 billion of data-center capital spending in 2026, with three of four hyperscalers punished by investors for the size of their commitments — is the context against which this price will eventually be judged[9]. Also absent so far: any published detail on German plants, jobs, or works-council terms, and any named regulator.
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asSLB's case is that it is not leaving its business — it is applying it. Moving heat around under pressure is what oilfield engineers have done for a century, and a data center is a heat problem with a different label[10]. The company argues it is buying real, installed, growing demand rather than a story: its own Data Center Solutions unit grew 63% year over year in the first half of 2026, and it expects that unit to pass $1 billion in annualized revenue by the end of 2026 and $2 billion by the end of 2027[8]. On price, SLB's argument is that 11 times EBITDA is a normal industrial multiple, not a technology multiple — and that after the $120 million of expected annual savings and cross-selling, the effective price falls to about 8.5 times[1]. Its deeper principle: oil demand is cyclical and politically contested, while the demand for cooling is a physical fact that does not care which way energy policy turns.
WhySLB wants to be valued as an industrial and infrastructure company rather than an oil-cycle company. Oil-services stocks trade at low multiples because their revenue swings with the crude price. Every dollar of revenue SLB moves into data centers is a dollar investors may pay more for[5][8]. The company also wants to keep deploying cash from a strong oil market into businesses that will still be growing when that market turns.
Impact on themSLB is spending $3.4 billion of cash and taking on $0.7 billion of debt at a moment when its shares were trading near a 12-month high[15]. If AI construction holds up, the combined data-center business is projected at $4.5 billion to $5 billion of revenue and $700 million to $800 million of adjusted EBITDA by 2028[6]. If it does not, SLB has bought a cyclical industrial asset at a full price, and the goodwill on its balance sheet becomes a future write-down risk.
Frames it asThe sellers' case is straightforward and rarely stated out loud: they bought an unglamorous German heat-exchange business, ran it, and are selling it into the strongest bid the market has produced in years. Apollo's own announcement calls Kelvion 'a global leader in cooling solutions for data centers and diversified industrials' and leads with the $4.1 billion total value, not the $3.4 billion cash figure[3]. That is not spin so much as the seller's accurate accounting: the enterprise they built is worth $4.1 billion, regardless of how the buyer splits it between cash and assumed debt.
WhyPrivate-equity funds exist to buy, improve, and exit. Selling at a moment when a strategic buyer is paying for AI exposure is the textbook execution of that model. The larger headline number also serves Apollo's fundraising: the size of a realized exit is what future investors look at[3].
Impact on themApollo and Triton convert an illiquid holding into cash on closing. Their exposure to whether the AI cooling boom continues past 2027 ends when the deal closes — a point worth holding next to SLB's, since the two parties are taking opposite sides of the same forecast.
Frames it asTheir argument is about physics, not sentiment. High-density AI server racks generate more heat per square foot than air cooling can remove at reasonable cost, so operators must move to liquid cooling — and liquid cooling requires heat exchangers, pumps and heat-rejection gear[10]. That demand is committed years ahead in construction contracts, not month to month. On this view Kelvion is a picks-and-shovels supplier: it gets paid whether or not any particular AI model succeeds, because the buildings are already being built. The best version of this case notes that Kelvion also serves ordinary industrial customers, so about half its revenue is not AI-dependent at all[1].
WhyInvestors and analysts positioned in energy-services and industrial stocks benefit when the market re-rates those companies as AI infrastructure plays[14][6].
Impact on themSLB shares traded higher on the announcement and touched a new 12-month high, according to market coverage that day[14][15].
Frames it asTheir case is not that AI is fake — it is that the spending is running ahead of the revenue, and that the risk is concentrated. Capital spending by the 14 largest public data-center operators is close to $750 billion in 2026, and three of the four hyperscalers lost market value after recent earnings calls specifically because investors balked at the size of those commitments[9]. Skeptics also point out that a sharp pullback need not be irrational to be painful: even an orderly correction in AI capital spending would hit suppliers hard and fast[9]. Applied here, the concrete worry is timing. SLB is paying about 11 times 2026 EBITDA for a company whose fastest-growing segment depends on orders placed by a handful of buyers, and the deal does not even close until the first half of 2027[1].
WhyShort sellers and cautious institutional investors gain from correctly calling a peak. But so do ordinary long-term holders who want management to avoid buying at the top of a cycle — the same criticism oil companies faced for acquisitions made at $100 crude.
Impact on themIf AI construction slows before or shortly after closing, SLB carries the asset and the debt. The sellers do not.
Frames it asKelvion has been passed between owners: it was carved out of German industrial heat-exchange manufacturing, held by private equity, and is now being sold to an American acquirer[12][3]. The interest here is continuity — plants, engineering jobs, and works-council agreements in Herne and other German sites. The strongest version of the concern is not anti-American: it is that a buyer paying a full price and promising $120 million in annual savings has, by its own arithmetic, committed to finding those savings somewhere[1].
WhyGerman works councils and unions typically seek binding site and headcount guarantees as a condition of ownership changes.
Impact on themNot yet determinable. Neither SLB nor Kelvion has published plant-level, headcount, or works-council commitments, and no German or EU regulatory review of the deal has been publicly named[1]. This is the single largest documented gap in the record as of September 1, 2026.
Like this article?
The Bias Ledger average rating 4
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| Bloomberg | U.S. center, financial | 2 | "SLB to Buy Data Center Cooling Firm Kelvion for $3.4 Billion" | Uses the buyer's cash figure in the headline rather than the $4.1 billion total value, and identifies Kelvion first as a 'data center cooling firm' rather than a century-old German industrial manufacturer. Both choices adopt SLB's framing of what was bought and for how much. The reporting itself is straight. |
| World Oil | U.S. energy-industry trade | 3 | Ran the deal under both figures on the same day — "SLB to acquire Kelvion for $3.4 billion in data center infrastructure expansion" and "SLB expands data center business with $4.1 billion Kelvion acquisition" | Publishing both numbers as separate headlines shows how unstable the price framing is, but the shared verb — 'expands' — is the company's own word. Coverage repeats SLB's growth projections without an outside estimate or a skeptic. |
| Data Center Knowledge | U.S. data-center industry trade | 3 | "SLB's $4.1B Kelvion Deal Expands AI Data Center Push" | Uses the larger total-value figure, which makes the deal look bigger for a readership that measures importance by deal size. Leads with the 2028 projection of $4.5–5 billion in combined revenue — a company forecast, presented as the story's payoff. |
| Benzinga | U.S. retail-investor financial | 4 | "SLB Eyes Rapid Data Center Growth Through $3.4 Billion Kelvion Deal" | 'Rapid growth' is SLB's projection stated as the frame of the story. Written for readers deciding whether to buy the stock, so the growth case gets the headline and the closing-risk and regulatory conditions get a clause near the bottom. |
| Fortune | U.S. center-left business | 5 | Earlier July 2026 profile: "Meet SLB: The $70 billion oil services giant poised to cash in on AI data centers and the post-Strait of Hormuz oil exploration boom" | 'Poised to cash in' is a forecast, not a report — and pairing AI upside with a war-driven oil boom builds an optimistic narrative for SLB months before this deal. Useful as evidence that the 'SLB as AI play' framing predates the acquisition and was already circulating when the deal was priced. |
| Tech Times | U.S. consumer tech | 7 | "SLB Acquires Kelvion for $4.1B: Air Cooling Fails AI Math, Oilfield Engineers Win" | The most editorialized framing in the set. 'Acquires' states as done a deal that has not closed and needs regulatory approval. 'Fails' and 'Win' declare an outcome the record does not support — no one has won anything until the first half of 2027 at the earliest. |
References
- SLB to Acquire Kelvion, Expanding its Role Across Data Center Infrastructure — SLB · Primary source — the acquiring company's own announcement; promotional by design
- SLB to Acquire Kelvion, Expanding its Role Across Data Center Infrastructure — Business Wire · Paid press-release distributor owned by Berkshire Hathaway; carries the company's text verbatim, no editing
- Apollo Funds Agree to Sell Kelvion, a Global Leader in Cooling Solutions for Data Centers and Diversified Industrials, to SLB for $4.1 billion — GlobeNewswire · Primary source — the selling private-equity firm's own release; frames the deal at its largest defensible value
- SLB to Buy Data Center Cooling Firm Kelvion for $3.4 Billion — Bloomberg · U.S. financial news owned by Bloomberg L.P.; serves institutional investors, market-centric framing
- SLB expands data center business with $4.1 billion Kelvion acquisition — World Oil · U.S. oil-and-gas industry trade publication; advertising-supported by the sector it covers
- SLB's $4.1B Kelvion Deal Expands AI Data Center Push — Data Center Knowledge · U.S. data-center industry trade publication (Informa); readership is the industry being covered
- SLB to acquire Kelvion for $3.4 billion from Apollo funds — Investing.com · Commercial retail-investor financial site; aggregates and summarizes company disclosures
- SLB: A New Growth Chapter Is Just Getting Started — Seeking Alpha · Contributor-written investment analysis; authors typically hold positions in the stocks they cover — treat as an interested bull case, not reporting
- AI Data Center Build Advances at Full Speed: Five Things to Know — BloombergNEF · Subscription energy-and-technology research arm of Bloomberg L.P.; sells data to industry and finance clients
- SLB Acquires Kelvion for $4.1B: Air Cooling Fails AI Math, Oilfield Engineers Win — Tech Times · U.S. consumer technology site; traffic-driven, heavily editorialized headlines
- Meet SLB: The $70 billion oil services giant poised to cash in on AI data centers and the post-Strait of Hormuz oil exploration boom — Fortune · U.S. business magazine, center-left editorially, executive-facing readership
- SLB to acquire Kelvion for $4.1bn — Heat Exchanger World · Netherlands-based industrial equipment trade publication; advertiser-funded by the heat-exchange sector
- SLB Limited/NV — Form 8-K, exhibit 99 (August 31, 2026) — U.S. Securities and Exchange Commission · Primary source — mandatory federal disclosure filing; content authored by the company but legally binding
- SLB Eyes Rapid Data Center Growth Through $3.4 Billion Kelvion Deal — Benzinga · U.S. retail-investor financial media; fast-turnaround, trade-oriented framing
- SLB (NYSE:SLB) Reaches New 12-Month High – Should You Buy? — The Cerbat Gem · Automated financial-content site (MarketBeat network); algorithmically generated stock summaries, not original reporting — used here only for the price-level data point