SLB 以逾 40 億美元收購 Kelvion，AI 資料中心散熱成能源服務新戰場

SLB 同意以約 34 億美元現金並承接 7 億美元債務收購 Kelvion，將熱管理能力納入快速成長的資料中心基礎設施業務。 交易的戰略意義不只在設備併購，而在液冷、換熱、模組化機房與能源管理能否被整合成可量測的交付單元。

SLB agreed to acquire Kelvion for about USD 3.4 billion in cash plus USD 0.7 billion of assumed debt, adding thermal management to its fast-growing data-centre infrastructure business. The strategic issue is not only an equipment acquisition, but whether liquid cooling, heat exchange, modular facilities and energy management can become measurable delivery units.

Kelvion 擁有百年熱交換與冷卻技術，近年把資料中心列為最大且成長最快的事業，同時服務氫能、碳捕捉、熱泵與工業市場。 Kelvion 的設備能力跨越冷卻塔、板式熱交換器與工業熱管理，併入 SLB 後可能影響資料中心設備選型、工程承攬與維運資料的責任邊界。

Kelvion has a century of heat-exchange and cooling experience. Data centres have become its largest and fastest-growing segment alongside hydrogen, carbon capture, heat pumps and industrial markets. Kelvion spans cooling towers, plate heat exchangers and industrial thermal management, so integration may change responsibility boundaries across design, procurement, commissioning and operations.

AI 算力增加使冷卻成為電力效率與新增容量的限制。交易規模與營收預測不等於實際節能，整合後仍需揭露設備效率、用水與全生命週期影響。 評估不能只看 PUE 單一數字，還要分開冷卻水來源、WUE、冷媒洩漏、熱回收、伺服器負載率與氣候條件，否則不同專案無法比較。

AI growth makes cooling a constraint on power efficiency and new capacity. Deal value and revenue forecasts do not prove savings, so equipment efficiency, water use and lifecycle impacts still require disclosure. PUE alone is insufficient. Review also needs water source and WUE, refrigerant leakage, heat recovery, server utilisation and climate conditions so projects can be compared.

SLB 計畫把 Kelvion 的熱交換與冷卻設備納入模組化資料中心方案，以系統整合、熱管理最佳化與更快部署回應客戶需求。 執行層需建立設計基線、設備序號、驗收條件、感測器校正、試運轉紀錄與維運責任，才能把系統整合承諾轉為可追蹤成果。

SLB plans to integrate Kelvion heat exchangers and cooling systems into modular data-centre offerings, using system integration and thermal optimisation to speed deployment. Implementation should preserve design baselines, equipment serials, acceptance criteria, sensor calibration, commissioning records and operational responsibility.

來源可核對的交易條件為約 34 億美元現金及 7 億美元承接債務；SLB 表示其資料中心業務 2024 至 2026 年營收年複合成長率預期逾 90%，年底累計交付容量可超過 2 GW。 目前已知資料主要證明交易條件與成長預期，尚未證明單一案場的節電、節水或範疇二減排成果，這些仍屬待驗證主張。

The source reports about USD 3.4 billion in cash and USD 0.7 billion of assumed debt. SLB expects data-centre revenue CAGR above 90 percent from 2024 to 2026 and cumulative delivered capacity above 2 GW by year-end. Available evidence establishes transaction terms and growth expectations, not site-level electricity, water or Scope 2 outcomes. Those remain unverified claims.

能源服務商正從油氣工程延伸到數位基礎設施，散熱設備的能效、冷媒與用水證據將直接影響資料中心的永續可信度。 散熱設備將從附屬工程變成 AI 基礎設施的產能與永續瓶頸，也會影響冷板、泵浦、熱交換器、控制器與測試服務的供應鏈。

Energy-service companies are moving into digital infrastructure, where evidence on cooling efficiency, refrigerants and water will shape sustainability credibility. Cooling is becoming both a capacity and sustainability constraint for AI infrastructure, affecting cold plates, pumps, exchangers, controls and testing services.

SNN 編輯分析：對台灣 AI 伺服器、散熱模組、泵浦、熱交換器與機房工程供應鏈而言，這筆交易提醒產業競爭將從零組件效率延伸到整套熱管理系統的可驗證交付。台灣業者若只提供型錄效率，可能無法回應雲端客戶對 PUE、WUE、冷媒、熱回收與實際負載條件的查核要求。揭露前證據基礎設施應至少保存設備 BOM 與序號、設計及測試基線、感測器校正、試運轉與驗收紀錄、運轉負載、逐時用電用水、維修更換及第三方測試結果。這些證據應能連回特定案場與版本，並明確區分模擬值、保證值與實測值。此段是台灣產業傳導的編輯解讀，不代表 SLB 或 Kelvion 已公布上述績效。

SNN editorial analysis: For Taiwan suppliers of AI servers, thermal modules, pumps, heat exchangers and data-centre engineering, the transaction suggests that competition is moving from component efficiency to verifiable system delivery. Catalogue values alone may not satisfy cloud customers reviewing PUE, WUE, refrigerants, heat recovery and actual load conditions. Pre-Disclosure Evidence Infrastructure should preserve equipment bills of materials and serials, design and test baselines, sensor calibration, commissioning and acceptance records, operating load, hourly electricity and water, maintenance replacements and third-party test results. Every object should resolve to a specific site and version, while simulated, guaranteed and measured values remain distinct. This is an editorial interpretation of Taiwan industry transmission, not evidence that SLB or Kelvion has reported those outcomes.

後續需追蹤交易核准、整合時程、Kelvion 訂單、實際 PUE 與用水表現，以及 SLB 是否把節能聲明轉為可稽核的專案數據。 編輯審查後續應把交易完成、訂單轉換與案場績效分開追蹤，並要求不同氣候區與負載情境下的可比測試。

Next checks include approvals, integration, Kelvion orders, measured PUE and water performance, and whether SLB turns efficiency claims into auditable project data. Editorial follow-up should separate transaction completion, order conversion and site performance, with comparable testing across climates and load profiles.