Immersion cooling: putting the servers in the fluid
Immersion cooling submerges entire servers in a dielectric fluid. It captures essentially all the heat, removes fans completely, and reaches densities nothing else matches — and it has stayed a specialist choice because it changes how every piece of equipment is handled and serviced.
In one sentence
Immersion cooling submerges computing equipment in a non-conductive fluid that carries heat away directly, either by circulating warm liquid or by boiling at the component surface and condensing above it.
In single-phase immersion, the fluid stays liquid and is circulated through a heat exchanger. In two-phase immersion, the fluid boils on the hot components and condenses on a coil above the bath, which is thermally very effective because it uses the latent heat of vaporisation. Both eliminate server fans and capture heat from every component, not only the ones with cold plates.
The obstacles are practical rather than thermal. Servers must be qualified for immersion, hard drives and some optical components do not tolerate it, maintenance means lifting a dripping server out of a bath, and the fluids are expensive. Two-phase fluids have also faced regulatory pressure over persistent chemicals, which has affected the technology's outlook.
How it works
Why it captures more heat
There is no air path and no component left uncooled: everything in the bath is in contact with fluid. That takes the fraction of heat handled by liquid close to the whole, removes fan energy entirely, and eliminates the hot spots that air-cooled designs work around.
What it costs operationally
Racks become tanks, which changes floor loading, hall layout and service procedures. Removing a server means lifting it out and letting it drain. Warranty terms must permit immersion. Every one of these is manageable and every one is a reason an operator with an existing air-cooled estate does not adopt it casually.
Where it is used
It has found real use where density is extreme, space is constrained, or the facility is purpose-built for a single workload — high-density computing sites and some purpose-built AI facilities. It has not displaced direct-to-chip cooling in mainstream builds, which reached adequate density with far less operational disruption.
What this depends on
3 of these are marked as a chokepoint: a handful of qualified suppliers, a multi-year lead time, or a single geography.
Supply chainChokepoint
Dielectric fluids
Specialised fluids with the right thermal and material compatibility properties, expensive and in some cases subject to regulatory pressure.
An ordinary chassis is not immersible: drives, some optical modules, labels and adhesives fail in the fluid, and the warranty has to permit it. The range of qualified hardware is narrow.
A tank exchanges its heat into the building's water loop like any other design, so immersion removes the fans but not the need to reject the heat outside.
Two-phase fluids sit inside the class of persistent chemicals being restricted, which makes the fluid a supply question rather than a technical one for the tank.
Who supplies this
What each company supplies at this step, and — where a public figure exists — its share of this specific market — with what that share measures, the period it covers and who published it. Some rows also show the company’s own reported revenue for the segment covering this step, which is a different thing: it says how much this business matters to that company, not how much of the market it holds. Not a ranking and not a recommendation.
Supplies servers built for immersion, which ordinary chassis are not.
What would change the picture
Whether regulation on persistent fluorinated fluids constrains two-phase deployment.
Whether any large operator standardises on immersion for new AI capacity.
Whether hardware vendors broaden immersion qualification across mainstream product lines.
Questions people ask about this
Does the fluid damage the electronics?
The fluids are chosen to be non-conductive and materially compatible with the components. The practical issues are with specific parts — spinning drives, some optical modules, certain labels and adhesives — which are removed or substituted when equipment is qualified for immersion.
Why hasn't it become standard?
Because direct-to-chip cooling reached the densities the market needed while keeping servers recognisable and serviceable in the normal way. Immersion offers more thermal headroom at the price of changing the physical operating model, and most operators have not yet needed that headroom.
Each page explains one technology in plain language, states what it depends on, and names companies by what they supply at that step. Company roles are described qualitatively and deliberately carry no market shares, revenue figures or rankings — those change faster than an explainer can, and a stale number is worse than none. Ticker links point at company pages on this site and are provided for reference only.
Nothing here is investment advice, a recommendation, or a forecast. A company named on a page about a technology is not thereby a good investment, and the chokepoints described are structural facts about supply chains rather than predictions about prices. Technology moves; where a page describes something as unresolved or in development, that was true when it was written.