SLB's $4.1 Billion Data Center Deal Is Sound — the Stock's AI Premium Isn't Free

The headline writes itself: the world's largest oilfield services company is spending $4.1 billion to ride the AI boom. SLB agreed in late August to buy Kelvion, a German maker of heat exchangers and cooling equipment, and the timing could not be better for a story. Cooling is one of the binding constraints on AI data centers, and investors have treated the move as a transformation: SLB is up nearly 50% year to date and trades around 27 times trailing earnings, a premium that none of its oil-service peers carry by comparison.

Before crediting the narrative, run the numbers the way you would for any capital decision: what did they actually buy, at what price, with whose money, and does it still pay the dividend?

What the $4.1 billion buys

Kelvion is not a stealth-AI startup. It is a century-old industrial company that makes thermal management systems — heat exchangers and cooling gear — for data centers, energy, and heavy industry. Since Apollo's funds took control at the start of 2026, the company has leaned hard into data centers, which have become its largest and fastest-growing end market, expected to bring in $1.2 billion to $1.3 billion of revenue this year.

The price is the interesting part. SLB is paying roughly 11 times Kelvion's estimated 2026 EBITDA before synergies, a multiple that drops to about 8.5 times once you count the $120 million in annual earnings SLB says it can generate from cost and revenue synergies within three years. That is a real industrial multiple for a real industrial asset, not a bubble price tag. Cooling genuinely is one of the biggest physical bottlenecks in the AI buildout, so this is not a fad multiple paid for a fad.

Where the money comes from, and what it costs shareholders

The structure is $3.4 billion in cash plus roughly $700 million of assumed debt. Against SLB's roughly $4.5 billion of trailing free cash flow, that is a large check to write in a single stroke. That being the case, the balance-sheet and income math holds. SLB says net debt to EBITDA stays inside its through-cycle target of up to 1.5 times, and it reaffirmed plans to return more than $4 billion to shareholders in 2026 and at least that much again in 2027.

For an income investor that is the load-bearing fact. SLB has paid a dividend for 24 consecutive years, runs a payout ratio near 50%, yields about 2%, and management says the deal adds to earnings and free cash flow per share in the first twelve months after closing. The deal is not even expected to close until the first half of 2027. This acquisition does not threaten the dividend; it is being funded largely from cash reserves without a financing raise.

The real test: scale, and what is already in the price

The engineering transfer is real, which is the part the AI label obscures. SLB builds modular, offsite-fabricated power, heat, and fluid systems for oil facilities, and a data center campus is, structurally, the same kind of large industrial project — just with servers instead of wellheads. Its data center business has been compounding at well over 90% a year, and with Kelvion combined, SLB expects more than $2 billion of data center revenue this year and targets $4.5 billion to $5 billion by 2028.

Here is the scale check that resets the story. That $2 billion is roughly 6% of SLB's annual revenue of around $36 billion. Even the ambitious 2028 target is only about a tenth of today's revenue base. SLB remains overwhelmingly an energy-services company with an AI segment bolted on, not an AI company with an energy division.

And the market has already moved ahead of that: the stock's year-to-date gain and its earnings multiple versus Halliburton's roughly 19 times put most of the AI enthusiasm into the price today, before the deal has even closed. The false narrative here is not that the acquisition is real — it is. The narrative worth testing is whether a segment that is still a single-digit share of revenue justifies the tech-style premium the market has quietly assigned the whole company. A dividend holder should come away reassured; a buyer chasing a stock that has already run on the AI story is paying for a premium the cash flow does not yet prove.