NextFin News - Blackstone Inc. has secured a first-lien term loan to fund its acquisition of a controlling stake in Flow Control Holdings, the maker of liquid-cooling components that sit at the heart of the artificial-intelligence data-center buildout. UBS Group AG has been mandated to provide the financing, according to a person familiar with the matter who asked not to be named discussing private information. The loan's size and terms were not disclosed.
The debt facility, arranged eight days after Blackstone agreed to buy Flow Control from Audax Private Equity, is the latest signal that lenders are willing to underwrite the industrial supply chain behind AI infrastructure — not just the hyperscale operators themselves. For Blackstone, the world's largest alternative asset manager, it also marks a shift from pure equity conviction to leveraged exposure in one of its highest-conviction themes. The deal shows how AI capital is migrating from the data-center landlords to the component makers that keep their servers from overheating.
The Deal: A Senior-Debt-Funded Pick-and-Shovel Bet
On September 10, 2026, Blackstone announced that private equity funds affiliated with Blackstone Capital Partners and Blackstone Energy Transition Partners had entered a definitive agreement to acquire Flow Control Holdings from Audax Private Equity. The Cincinnati, Ohio-based company manufactures highly engineered flow-control components used in coolant distribution units, in-row manifolds and secondary fluid networks — the plumbing that keeps AI server racks from overheating. Audax will retain a minority equity stake and continue partnering with Blackstone and Flow Control's management after the transaction, which is expected to close in the fourth quarter.
Financial terms of the acquisition were not disclosed. But the financing now in place tells part of the story: a first-lien term loan — the senior-most debt in the capital structure, repaid ahead of other creditors — arranged by UBS, which was already serving as a financial advisor to Blackstone on the equity side of the transaction alongside William Blair and Goldman Sachs. The same bank advising on the buy is now funding it, a common pattern in sponsor deals where the lender earns both advisory fees and interest income.
Flow Control's customer base spans original equipment manufacturers and hyperscalers building liquid-cooled data centers, plus sanitary and high-purity applications in the food, beverage and pharmaceutical markets. That diversification matters: it gives the business cash-flow stability beyond the AI cycle, even as the data-center cooling segment supplies the growth narrative that justified Blackstone's bid. The company's components are specified into systems rather than sold on a spot market, which makes revenue more predictable — a key requirement for a business carrying acquisition debt.
"We believe that FCH has enormous tailwinds for continued growth as the latest generations of high-performance AI infrastructure increasingly rely on liquid cooling technology for significantly improved energy and chip efficiency. We look forward to partnering with Scott and the entire FCH management team, as well as Audax, to invest in the company's capacity expansion plan, support unprecedented liquid cooling demand, and further accelerate growth behind one of Blackstone's highest-conviction investment themes – the immense buildout of AI infrastructure."
That statement came from Bilal Khan, a senior managing director at Blackstone, and Mark Zhu, a managing director at the firm. Scott Kerns, Flow Control's chief executive, said the company had spent the past four years building its position in data-center cooling, backed by Audax's capital for new production and distribution facilities and a string of 10 acquisitions. The buy-and-build track record is precisely what makes the company a leveraged-buyout candidate — and what makes the new debt load defensible. Audax Private Equity partner Don Bramley called the sale "a text-book example of Audax's ability to build strategic assets through disciplined execution of the Audax Value Agenda and Buy & Build approach."
Why the Instrument Matters More Than the Price
The most important detail in this transaction is not the undisclosed price — it is the instrument. A first-lien term loan is cheap, senior, covenant-protected capital. By using it rather than funding the acquisition entirely from its equity funds, Blackstone is doing what private equity does best: layering low-cost debt beneath an asset whose cash flows are expected to grow faster than the cost of that debt. The spread between the two is the return, and the first-lien position means Blackstone's lenders get paid before any other creditor if the business stumbles.
But the choice of collateral is the real tell. Blackstone is not buying another data-center operator. It already owns more than $150 billion of data-center assets globally, including QTS and AirTrunk, and QTS's leased megawatts have grown roughly 14-fold since Blackstone took it private in 2021. Instead, it is buying the component supplier — the pick-and-shovel play in the gold rush. When every hyperscaler is racing to install liquid cooling to handle denser AI racks, the supplier of fittings, manifolds and distribution units collects revenue from the entire industry, not just one operator's utilization rate.
That logic is backed by market sizing. The data-center liquid-cooling market is projected to expand from $6.6 billion in 2026 to $38.4 billion by 2033, a compound annual growth rate of 28.7%, as direct-to-chip and immersion cooling displace air cooling in high-density facilities. Liquid cooling is not an optional efficiency upgrade; it is becoming a physical requirement. Air simply cannot remove enough heat from racks packed with the latest AI accelerators, and power-usage-effectiveness targets make the thermal equation a board-level issue rather than a facilities one.
The financing also arrives inside a much larger debt boom. At least $334.5 billion of bonds and loans tied to AI infrastructure had been issued so far in 2026, compared with $185.5 billion across all of 2025 — an increase of roughly 80% year over year. Blackstone's own AirTrunk platform is finalizing a A$4.3 billion (roughly $3 billion) construction loan for its SYD3 hyperscale facility in Australia and completing an approximately $2.3 billion loan for a project in Johor, Malaysia. In May, Blackstone's data-center vehicle, Blackstone Digital Infrastructure Trust, priced an initial public offering of 87.5 million shares at $20 each — up to $2.0 billion of gross proceeds if the over-allotment is exercised — and began trading on the New York Stock Exchange under the ticker BXDC. The trust targets newly built data centers valued between $250 million and $1.5 billion, leased to investment-grade hyperscalers, with expected yields of 5.75% to 7% and annual rent escalators of 2% to 3%.
Put together, the Flow Control loan is one piece of a coordinated capital stack: public equity (BXDC), private equity (the Blackstone funds), and now senior bank debt (UBS) all flowing into the same theme at different points in the value chain. The breadth of that stack is the point — it shows how deeply institutional capital has committed to AI infrastructure, and how many different ways there now are to express the same conviction.
Cyclical or Structural: This Is a Regime Shift, Not a Wave
The critical question for any investor in this trade is whether liquid-cooling demand is cyclical — a mean-reverting surge that will fade — or structural, a regime shift that will not reverse on its own. The evidence points decisively to structural, for three reasons.
First, the driver is physics, not sentiment. Chip power density has crossed a threshold where air cooling is no longer technically sufficient for the highest-performance AI workloads. Once a data hall is retrofitted or built for liquid cooling — with coolant distribution units, manifolds and secondary fluid networks installed — the infrastructure does not get removed when the next GPU cycle arrives. It is a capital-intensive, sticky installed base, and replacement demand compounds on top of new-build demand.
Second, the installed-base math is only beginning. The market is projected to grow nearly six-fold over seven years. A cyclical boom does not sustain a 28.7% compound growth rate across an entire decade; that is the signature of a technology transition still in its early innings. For comparison, the air-cooling era lasted for decades; the liquid transition is measured in quarters.
Third, the customer concentration cuts the other way. Flow Control sells to OEMs and hyperscalers whose own capex is enormous and multi-year. Hyperscaler capital expenditure is not a discretionary line item that gets cut at the first sign of a slowdown — it is the infrastructure spend behind the AI revenue model itself. As long as AI model training and inference demand keeps growing, the cooling underneath it must grow too.
That said, the financing is not risk-free. The debt is cyclical in one important sense: it was priced into a credit market that has been unusually accommodating to AI-linked borrowers. If rate expectations shift or if lenders begin to discriminate between core data-center assets and peripheral suppliers, the refinancing environment in three to five years could look very different from today. The structural demand for cooling does not guarantee a benign cost of capital. A first-lien loan is senior, but seniority does not protect against a borrower whose revenues fall faster than its debt service.
The Second-Order Trade: Leverage Is Spreading Down the Supply Chain
The first-order read of this news is simple: Blackstone is bullish on liquid cooling. The second-order implication is more consequential. Debt is now reaching the industrial suppliers at the edge of the AI buildout — the companies that make the components, not just the ones that own the buildings.
That matters because it changes the risk distribution of the AI infrastructure boom. When only data-center owners borrowed, the leverage sat on long-lived real estate with investment-grade tenants. As financing extends to component makers, leverage moves onto shorter-duration industrial cash flows that are more exposed to order volatility. A slowdown in hyperscaler orders would hit a supplier's revenue faster than it would hit a landlord's rent roll, because leases are sticky while component orders are not. The first-lien structure protects lenders in a liquidation, but it cannot prevent a revenue shock from squeezing the borrower's margins first.
There is also a competitive dynamic worth watching. Liquid cooling is attracting industrial incumbents — companies like Vertiv, nVent and Parker Hannifin have deep engineering benches and global distribution. Flow Control's advantage is specialization and speed, built through a decade of acquisitions. Blackstone's plan to fund capacity expansion is the right response: in a supply-constrained market, the winner is often the company that can scale production fastest, not the one with the lowest unit cost today.
The counter-thesis is straightforward and deserves weight. Hyperscaler capital expenditure is not infinitely elastic. If AI monetization lags deployment — if the revenue from AI services does not justify the buildout — capex guidance could contract, and component suppliers would be the first to feel it. Flow Control's food, beverage and pharmaceutical businesses provide a cushion, but they also dilute the pure-play AI exposure that justified the premium. Integration risk is real too: a company that grew through 10 acquisitions under one owner must now be integrated under another, with new debt service on top. And Blackstone is not the only sponsor seeing the opportunity; competition for cooling assets could push entry multiples to levels that leave little room for error.
The signal that would falsify the structural-demand thesis is specific and observable: if two or more major hyperscalers cut their 2027 data-center capex guidance by double-digit percentages, or if Flow Control's data-center revenue growth decelerates materially in the first four quarters after closing, the "unprecedented demand" narrative would be in trouble. Until then, the physics of chip density and the backlog of AI facilities under construction support the structural call.
What Comes Next: Three Scenarios
In the short term, the story is execution: closing the transaction in the fourth quarter and announcing the capacity-expansion plan that Blackstone promised. The market will watch for the first post-closing orders and any disclosure of the purchase price, which remains undisclosed. The UBS-mandated loan suggests the financing leg is complete; the remaining work is operational.
Over the medium term, the key variable is the credit market. The UBS-mandated first-lien loan is a data point, not a trend — the question is whether more banks follow with similar financing for AI supply-chain assets, or whether this deal proves to be an outlier. A widening of similar mandates would confirm that lenders see the supply chain as bankable collateral; a drought would suggest the Flow Control loan was an exception granted to a marquee sponsor with a $1.3 trillion balance sheet behind it.
In the long term, the structural thesis stands or falls on installed-base persistence. If liquid cooling becomes the default architecture for AI data centers — as the power-density trajectory suggests — Flow Control's components become recurring, spec-in revenue rather than one-off project work. That is the outcome Blackstone is underwriting.
Three scenarios frame the outlook. In the base case, hyperscaler capex stays elevated, liquid-cooling penetration keeps rising, and Blackstone exits or refinances the position at a multiple expansion supported by earnings growth. In the upside case, a hyperscaler capex acceleration or a supply shortage sends component pricing higher, and Flow Control becomes a platform for further add-on acquisitions under Blackstone's ownership. In the downside case, AI monetization disappoints, capex guidance contracts, and the senior debt — while first in line — still faces a smaller, less profitable borrower.
The takeaway is this: Blackstone is not betting that one data-center operator will win. It is betting that every operator will need the same plumbing — and it is using senior debt to make that bet cheaper. The loan is small news on its own; the pattern it confirms is not. When the pick-and-shovel sellers start carrying leverage, the gold rush has moved past the early prospectors and into the hands of the banks.
Explore more exclusive insights at nextfin.ai.
