Datacenter Cooling — What Matters This Week (May 2026)

1. Top 3 Things Happening (User View)

1. **Munters FY2025 year-end report (reported Q1 2026) lands full-year net sales above the ~SEK 15.5B FY2024 base, with Data Center Technologies named the primary growth engine; Q3 2025 already showed sustained double-digit DCT demand as customers shifted to high-density.** This is the GB200/GB300 inflection arriving in audited financials: Products module says racks crossed ~100–130kW in 2025–2026, breaking air's envelope — Munters is the first cooling vendor where that shows up as a reported P&L line, not a forecast.

• Why it matters to: COOLING-OPS-DIR-01 (air→liquid transition at scale) and VCs scanning public comps.
• Action: Re-baseline vendor shortlists against Munters' DCT (~45% of group) as the public proxy for liquid demand — stop using private startups as the market read.

1. **Incumbent OEMs are converting legacy manufacturing into owned CDU product lines: nVent posting Manufacturing Engineer, Cooling Enclosures (eeb7a34c, 2026-04-30) plus core CDU reqs (d834c2ed / 13a32eaf, 2026-03/04), while CoolIT opened Technical Sales Manager, Hyperscale Accounts and a dedicated CDU mechanical engineer (CDU module = owned line, not just cold plates).** Why now: NVIDIA reference designs made direct-to-chip + CDU the *default* across 2025–2026 (Products module) — that pulled enclosure/manifold makers into the rack.

• Why it matters to: jobseekers and COOLING-OCP-CONTRIB-01 tracking reference-design adoption across members.
• Action: Treat CDU mechanical + hyperscale TSM reqs as the entry door this cycle; target nVent/CoolIT before headcount bands formalize.

1. **Cooling hiring went from 0 tracked roles to 30 net-new in one 90d window, 8 of them "liquid cooling design" across 5 companies — but 100% of postings carry unspecified seniority, so the org ladder is structurally invisible.** This is a labor category created inside a single hardware cycle by the power-density jump, not a gradual ramp.

• Why it matters to: COOLING-OPS-DIR-01 (staffing liquid-maintenance is a different skill set) and jobseekers.
• Action: Don't read seniority from titles this quarter; benchmark comp/scope via the hiring company's stack-layer and CDU ownership instead.

2. Strategic Implications (Our View)

1. **Every L1 "cooling vendor" in coverage is an incumbent OEM mid-pivot, not a VC-backed newcomer.** Modine (auto-thermal), nVent (connection/protection), Asetek (consumer AIO), Munters (air handling) are all repurposing manufacturing scale.

• Evidence: Market module — "Modine guided to a ~$1B data-center sales run-rate over FY2025–FY2027 versus a sub-$200M base ~3 fiscal years earlier [signal 1381382e]."
• Implication for next 30 days: Stop framing this Topic as a startup race; reframe newsletter + persona content around incumbent-capture, which is the non-consensus read worth distributing.

1. **Hiring is loud while business signals are silent — a hire-ahead-of-revenue phase.** Modine, nVent, Iceotope, Asetek all show "no material signals (90d)"; only Munters and CoolIT carry material business signals.

• Evidence: Company modules — Iceotope/Modine/nVent/Asetek "Recent Business Signals (90d): _(no material signals)_".
• Implication for next 30 days: Use hiring as the leading indicator and explicitly tell readers business signals lag ~2–3 quarters; weight CDU/liquid-design req velocity over press releases.

1. **Differentiation has collapsed to one axis: qualified into an OEM server line today vs. still hiring core CDU engineers vs. shipping research as if it were product.** Modine's two-phase work is research, not a shippable line.

• Evidence: Products module — competitive question is "who is qualified into OEM server lines today versus who is still hiring core CDU engineers ... and who is shipping research, not product (Modine two-phase, 51ff86d1)."
• Implication for next 30 days: Add a binary "OEM-qualified?" flag to vendor scoring; it predicts more than fit score.

3. Newsletter Issue Spotlight

• Lead story: The cooling "winners" are 70-year-old incumbent OEMs in disguise, not startups (Implication #1 + Thing #1/#2).
• 3-second hook: Every datacenter-cooling 'winner' is a decades-old parts maker in disguise.
• Image direction: V16 isometric 3D — a vintage automotive radiator/heat-exchanger on the left morphing mid-frame into a glowing GB300 liquid-cooled CDU rack on the right, single light source, clean studio base, no text.
• Subject line: Cooling's winners aren't startups, they're OEMs

4. Reverse-Hype Watch

• Overhyped right now: Two-phase / advanced cooling R&D narratives. Modine's two-phase work (signal 51ff86d1) is being read as a product when the Products module explicitly classes it as "shipping research, not product" — and Modine shows zero material business signals in 90d.
• Underrated right now: Munters as effectively a datacenter company. Data Center Technologies is ~45% of SEK 15.5B group sales and the FY2025 growth engine (Munters module), yet it's still filed mentally under "air handling." The public proxy for liquid demand is hiding in plain sight.

5. Pattern Library Updates

• Success pattern observed: Incumbents that repurpose high-volume manufacturing into datacenter cooling hardware scale revenue fastest — Modine moved from a sub-$200M base to a guided ~$1B data-center run-rate (signal 1381382e); Munters DCT is the FY2025 growth engine.
• Failure pattern observed: Scoring a vendor on technology/R&D claims (Modine two-phase, 51ff86d1) or pre-revenue pivot intent (Asetek, no live deal signals) instead of OEM-qualification status — it overstates readiness by 2–3 quarters. Avoidance: gate vendor fit scores behind a binary "qualified into an OEM server line today?" check.