The 2026 immersion cooling supply chain is dominated by single-phase architectures, with synthetic hydrocarbons commanding 55.0% of the fluid-chemistry segment and new-build data centers representing 41.0% of the deployment-model mix [S3].
Dielectric fluid availability, CDU/pump sourcing, and tank-fabrication capacity are the three pinch points that determine lead time, while 3M's 2022 exit from its Novec fluorocarbon line permanently reshaped the two-phase supply curve and pushed commercial buyers toward single-phase designs [S7].
Synthetic Hydrocarbons account for 55.0% of the Fluid Chemistry segment in the immersion cooling market in 2026, according to Future Market Insights analysis.
Synthetic hydrocarbons lead the 2026 fluid-chemistry mix at a 55.0% share, displacing legacy mineral-oil and fluorocarbon options as the default dielectric for new commercial deployments [S3]. The dielectric fluid tier is the most volatile link in the immersion cooling supply chain, with availability and price swings on highly refined mineral oils, synthetic esters, and fluorocarbons flowing directly through to total system cost [S4].
For 2026 sourcing, buyers should benchmark fluid specifications against material compatibility (elastomer swell, copper-tarnish behaviour), dielectric breakdown voltage (typically >30 kV per ASTM D877 for qualified fluids), and supplier disclosure of base-oil source, since re-refined versus virgin stocks behave differently on long-term oxidation stability. Two-phase fluids remain a constrained specialty after 3M's 2022 Novec exit collapsed that supply curve, leaving the architecture active at select hyperscaler research sites rather than as a mainstream procurement option [S7].
Single-Phase vs Two-Phase: 82% of Non-Hyperscale Installs Go Single-Phase
Single-phase immersion cooling accounts for more than 82% of non-hyperscale 2026 commercial installs, a share driven less by peak thermal performance than by serviceability and supply-chain resilience [S5]. Two-phase designs submerge servers in a low-boiling-point fluid and transfer heat through latent heat of vaporization, delivering strong theoretical thermal performance; the practical supply chain for those fluids, however, has not recovered since 3M announced the planned discontinuation of its Novec family in 2022 [S7].
The decision matrix for 2026 buyers can be summarized against four criteria:
Architecture selection in 2026 is rarely a pure thermal-engineering decision: buyers evaluating industrial UPS adjacencies for AI-training halls should weight fluid-supply continuity, condensate management, and service-procedure maturity alongside raw heat-transfer performance, since two-phase's theoretical advantage evaporates when qualified fluid inventory is unavailable at the project's energization date.
System CAPEX Stack: Tank, CDU, Fluid First-Fill, Heat Exchanger
Immersion cooling CAPEX in 2026 is a four-line stack: tank, Coolant Distribution Unit (CDU), dielectric fluid first-fill, and heat-exchanger/pump skids, with TCO models typically assuming a 5-year lifecycle at a 4% discount rate against US industrial electricity of USD 0.08-0.15/kWh [S6]. Air cooling's comparable CAPEX line — CRAC/CRAH units, chilled-water plant, and raised floor — is increasingly disqualified at densities above ~50 kW per rack, which is the operating band where immersion cooling procurement now concentrates [S6].
For practical specification work, the fluid first-fill line is the variable that breaks fixed budgets. A 1 MW immersion deployment can require tens of thousands of litres of qualified dielectric, and Shell's June 2025 launch of DLC Fluid S3 (a propylene-glycol-based direct liquid cooling fluid rated for freeze protection down to -10°C, >6-year life, and OCP PG25 compliance) shows how fluid suppliers are repositioning to capture the data-center adjacency that was historically held by immersion-only vendors [S2]. This is also why supply-chain managers handling switching power supply deployments for adjacent IT loads should treat fluid sourcing as a critical-path item rather than a procurement tail.
Dielectric Fluid Sourcing: Novec Exit, 3M Aftermath, and PG25 Compliance
3M's 2022 announcement of the planned discontinuation of its Novec fluorocarbon family remains the single largest structural shock in the immersion cooling fluid market, and its effects are still propagating through 2026 procurement cycles [S7]. Two-phase immersion designs that depended on Novec-engineered fluids have either requalified to alternative fluorocarbons (with associated dielectric-breakdown and GWP re-testing) or migrated to single-phase architectures where the supply base is broader [S7].
Buyers writing RFQs in mid-2026 should require suppliers to disclose base-chemistry family, dielectric breakdown voltage per ASTM D877, kinematic viscosity at 25°C and 40°C, flash point (COC), and explicit OCP PG25 or PG25-equivalent compliance for any fluid pitched at AI/HPC workloads. Shell's DLC Fluid S3 launch is a useful 2025 reference data point: a propylene-glycol chemistry, fluorescent-green leak-trace dye, claimed PUE improvement of up to 27% versus air cooling, and a stated service life above six years [S2]. Comparable disclosure depth from incumbents like Shell plc, which holds an estimated 11.0% share of the broader immersion cooling market, is the baseline that specialist single-phase vendors are now expected to match [S3].
Equipment Tier: Tanks, CDUs, Heat Exchangers, Pumps, Monitoring
The equipment supply chain splits cleanly into five sub-tiers: immersion cooling tanks, coolant, CDUs, heat exchangers, and pumps/fluid-circulation plus monitoring-and-control systems, with each tier carrying distinct vendor concentration and lead-time profiles [S8]. Tank fabrication is typically the longest-lead mechanical item (custom stainless or coated carbon-steel vessels with welded headers), while CDUs and plate-frame heat exchangers are largely off-the-shelf skids with 8-16 week lead times in 2026 [S8].
Pumps and monitoring-and-control stacks lean on the same suppliers who already serve dc power supply and process-control channels, so buyers who already run those categories inside a hyperscale or colocation procurement function can usually fold immersion cooling into existing vendor-master agreements rather than build a parallel sourcing organization. Modine's January 2024 acquisition of TMG Core illustrates how thermal-management OEMs are vertically integrating immersion IP into established HVAC and chiller portfolios, a structural shift that compresses the independent-vendor field in the CDU and heat-exchanger tiers [S2].
Regional Sourcing: North America Leads, India at 17.3% Growth
North America remains the largest regional market for immersion cooling in 2026, anchored by the densest concentration of AI-training and hyperscale build-outs, while India is the fastest-growing country market at 17.3% CAGR as data-center infrastructure programmes scale [S3]. Shell plc leads the competitive field at an estimated 11.0% market share, supported by production scale and established supply agreements, with GRC (Green Revolution Cooling), Modine (post-TMG Core), and a long tail of specialist single-phase vendors filling the rest of the field [S1][S3].
For global buyers, the practical 2026 sourcing pattern splits into three lanes: hyperscaler direct-procurement contracts (North America, Western Europe) where Shell/Modine/GRC-class vendors are the typical bidders; India and Southeast Asia where regional integrators package imported fluids with locally fabricated tanks; and edge/micro-site deployments where pre-engineered containerized immersion units dominate because they bypass the on-site tank-fabrication lead time. The 2023 GRC / Dell Technologies / DCV Industries partnership for Middle East & Africa deployment is a working template for that third lane [S1].
Procurement Signals: Lead Times, Margin Compression, Service Reliability
Pricing pressure from institutional buyers and procurement consolidation is compressing 2026 immersion-cooling margins, which shifts competitive differentiation toward product performance, service reliability, and total cost of ownership rather than headline thermal specs [S3]. Two operational signals to track: dielectric fluid lead times (still the highest-variance line in the BOM) and CDU retrofit-cycle compatibility with existing chilled-water plants, since brownfield sites without a chilled-water loop face materially higher integration cost [S6].
Trackable next nodes for Q3-Q4 2026 include any new PG25-compliant propylene-glycol or synthetic-hydrocarbon fluid launches following Shell's DLC Fluid S3 template, two-phase fluid requalification announcements at hyperscaler scale, and the next Modine-TMG Core product release cadence, since those three signals together determine whether single-phase's >82% non-hyperscale share holds or two-phase recovers ground in commercial procurement [S2][S5]. Buyers working through the same procurement organizations that source roller chain and other industrial MRO categories can usually extend the immersion sourcing workflow without standing up a new vendor-qualification cycle, provided fluid and CDU tiers are tracked as critical-path items from day one.
See also our earlier report, Vibration Meter Price 2026: Cost Drivers, Tier Comparison, and Sourcing Specs.