Enterprise IT Hardware : Building Sustainable and Adaptive Workplace Systems in 2026

Enterprise IT hardware stands at a turning point. Sustainability has moved from a peripheral KPI to a core design and procurement constraint, driven by the twin realities of embodied carbon (from manufacturing servers, chips, storage) and operational carbon (from electricity and cooling to run them). Hyperscalers and data center operators are quantifying and tackling embodied carbon with new methodologies. Meta, for instance, now estimates and tracks emissions down to the component level and is pairing this with aggressive operational decarbonization via energy efficiency and carbonfree power procurement. Workload growth, especially AI training and inference keeps surging while grids face capacity constraints and regulators raise the bar (with initiatives like the EU’s Ecodesign for Sustainable Products Regulation (ESPR), RighttoRepair Directive, and climate disclosure frameworks). Enterprises must simultaneously manage rising demand with tighter environmental accountability and cost pressures.  

For procurement leaders, the stakes are high: total cost of ownership now depends equally on energy, reuse, and disclosure readiness. When data centers run on renewable energy, most of their remaining emissions come from making the hardware, sometimes as much as 50–80%. So, companies need to focus on designing hardware that lasts longer and can be reused or recycled. 

Four key trends transforming enterprise IT hardware strategy in 2026  

Trend #1: Circular and modular hardware becomes a core procurement focus 

Organizations are shifting from linear refresh cycles toward circular and modular models – designing for reuse, refurbishment, and parts harvest, and extending lifespan via memory disaggregation (e.g., CXL), component reuse, and service‑based leasing. This mitigates embodied carbon and supply risk while meeting emerging regulations on durability and repairability. 

Key drivers 

• Regulatory pressure: The EU’s ESPR (in force July 2024) expands Ecodesign beyond energy to durability, repairability, recyclability; the Right‑to‑Repair Directive makes repair obligations enforceable. The member states are in the process of transposing the latter into national law by July 2026, while the first specific ESPR product rules (including for ICT products) are being finalized for application in the coming years. 

• Macro risks and costs: Volatile supply chains and rising capital costs make recovered components and secondary markets financially attractive. Leasing/PaaS models now achieve >90% return rates, securing material streams and reducing Scope 3. 

• Corporate climate commitments: As operational emissions fall via renewable procurement, embodied carbon becomes the dominant share, requiring action in server design, memory, and storage reuse. 

Evidence and examples

Microsoft Circular Centers are reusing/recycling 90% of decommissioned cloud hardware; >3.2 million components were reused in FY24, creating measurable value recovery

Dell Asset Recovery Services standardizes global disposition, harvesting parts and re-marketing whole devices to offset new investments; case studies show budget and sustainability gains

Circular storage: A multi‑partner study found 87% of drives are suitable for secure erasure and reuse, revealing a large, untapped reservoir of embodied carbon savings

Implications for category managers

• Build RFP criteria around design for return (field‑replaceable modules, standardized fasteners), component reuse pathways (CXL‑enabled memory pools; certified SSD/HDD refurbishment), and take‑back and circular performance metrics (percent reuse, value recovery, embodied carbon per SKU).

• Embed clauses that require Digital Product Passport metadata for sustainability/repair to anticipate ESPR delegated acts.

Trend #2: High‑density computing meets holistic energy and thermal optimization

High-end racks (specialized physical cabinets and accompanying infrastructure designed to support power-hungry, high-performance computing (HPC) environments) are reaching much higher levels, with some AI-focused racks exceeding 132 kW, and projections suggesting densities could reach 240 kW in the near future, making air cooling insufficient and elevating energy/water constraints. Liquid cooling (direct‑to‑chip, immersion) and AI‑optimized operations are moving mainstream to maintain Power Usage Effectiveness (PUE) targets and grid reliability.

Key drivers

• Workload growth: Global data center electricity consumption reached approximately 448-460 TWh in 2025 and is on track to more than double by 2030, driven by the substantial growth of AI training and inference. Some estimates suggest consumption could reach nearly 1,000 TWh by 2030 without significant efficiency gains, with AI servers representing a substantial portion of that increase.

• Grid constraints & reliability: Tier‑1 markets face power scarcity; operators are moving to hourly 24/7 carbon‑free energy matching, microgrids, and waste‑heat recovery to stabilize operations and lower Scope 2 emissions.

• Thermal standards & technology: Accelerators and next‑gen CPUs/GPUs necessitate liquid cooling to cut cooling energy and meet thermal envelopes. Analyses show significant PUE improvements when introducing liquid cooling.

Evidence and examples

Liquid cooling dominance: Liquid cooling adoption in new data center deployments is accelerating rapidly, moving from roughly 10% market share in 2024 toward over 20% in 2025, essential for high-density AI servers and achieving aggressive PUE targets below 1.2.

Operator benchmarks: Leading hyperscalers such as Google report exceptional PUEs, with a fleet-wide average of 1.09, far outpacing the industry average of ~1.56, demonstrating superior operational efficiency and integrated thermal strategies.

Holistic guidance and renewables: The industry emphasizes comprehensive approaches, including workload optimization and advanced siting strategies, to manage growth. Hyperscalers remain the primary drivers of corporate Power Purchase Agreements (PPAs), accelerating commitments to 24/7 carbon-free energy to meet rising demand sustainably.

Implications for category managers 

Prioritize solutions that 

• Support liquid cooling (cold‑plate ready racks, compatible manifolds)  

• Include telemetry for dynamic cooling setpoints 

• Enable energy‑aware workload placement (following renewables by time/location). Require vendors to disclose thermal performance with PUE/WUE impacts and integration options for heat reuse or storage. 

Trend #3: Computing architecture pivots to efficiency 

To sustain performance without proportional increases in power consumption, organizations are adopting more efficient computing architectures. These include ARM-based CPUs (central processing units) for selected workloads, DPUs (data processing units), also known as SmartNICs, to offload networking, storage, and security tasks from host CPUs, and CXL-enabled (Compute Express Link) memory expansion to pool and reuse DRAM (dynamic random-access memory) more effectively. 

Together, these approaches reduce CPU overhead, lower power consumption, and enable more modular, right-sized infrastructure configurations. 

Key drivers 

• Rapid growth in core counts and heterogeneous CPU designs: Processor roadmaps show 128–192+ cores per socket, with AMD currently leading among x86 (traditional server) CPU architectures. The widespread adoption of mixed Performance (P-) cores and Efficient (E-) cores allows better performance-per-watt optimization across different workloads, particularly AI-driven tasks. 

• Economics of network and infrastructure offload: Adoption of DPUs (data processing units), also known as SmartNICs (smart network interface cards), is accelerating, with the market growing at a compound annual growth rate (CAGR) of over 13%. Platforms such as NVIDIA BlueField offload functions like TCP/IP networking, encryption, storage protocols, and microservices from host CPUs – freeing processing capacity while improving energy efficiency and security. 

• Memory bottlenecks and improved reuse: The finalized CXL (Compute Express Link) 3.0 standard – with products expected in 2026 – lays the groundwork for full memory disaggregation and coherent memory sharing across multiple hosts. This shift is expected to significantly reduce memory overprovisioning and increase reuse of DRAM (dynamic random-access memory), helping lower both infrastructure costs and embodied carbon emissions. 

Evidence and examples

BlueField DPUs: Studies and solution briefs indicate reduced host CPU utilization and power efficiency gains by offloading networking and security tasks; research prototypes show substantial throughput improvements and lower CPU cycles for small‑packet workloads.

SCARIF & embodied modeling: New academic tools such as SCARIF (Server Carbon Assessment and Reporting Impact Framework) are enabling more precise measurement of embodied carbon across different server configurations. These models allow organizations to compare architectural options not only on performance per dollar, but increasingly on performance per kilogram of CO₂ equivalent (kgCO₂e).

Implications for category managers 

• Build efficiency and sustainability metrics into sourcing requirements: Ask suppliers to include performance-per-watt and carbon impact (kgCO₂e) in bids. Request server options with energy-efficient designs – such as DPUs/SmartNICs, CXL-enabled memory, and a mix of ARM and x86 architectures – matched to how the systems will actually be used. 

• Factor sustainability into comparisons: Where possible, reuse existing memory and storage components, and compare suppliers on embodied carbon across server configurations, using consistent methods that account for differences in manufacturing location and timing. 

Trend #4: Disclosure readiness and climate governance  

Regulatory and investor scrutiny is rising even as U.S. federal timelines evolve. Companies face EU CSRD, California SB 253/261 momentum, and evolving SBTi criteria (Science Based Targets Initiative). Procurement must ensure suppliers can support reliable Scope 1–3 reporting, 24/7 CFE tracking, and TCFD‑aligned risk governance.   

Key drivers 

• SEC climate rule dynamics: The SEC adopted climate disclosure rules (March 2024) requiring material Scope 1–2 for certain filers, though litigation led to a stay and subsequent rollback signals; market expectations remain elevated. 

• Global alignment: The ISSB (International Sustainability Standards Board) standards have effectively superseded the TCFD (Task Force on Climate-related Financial Disclosures) framework, with the TCFD disbanding and the IFRS Foundation taking over monitoring. Over 35 jurisdictions are now adopting or aligning with the ISSB’s IFRS S1 and S2 standards, making them the de facto global baseline for climate risk reporting, while leading firms continue to conduct scenario analyses to meet investor demands. 

• SBTi updates: The SBTi is developing its Corporate Net-Zero Standard Version 2.0 (V2), with the second draft released for consultation in late 2025 and an expected final version in 2026. This process is focused on enhancing accountability, refining the role of carbon removals, and may include more specific guidance on Scope 2 emissions accounting. 

Implications for category managers

• Companies should embed stringent climate performance clauses into supplier contracts to meet evolving global regulations (ISSB, ESRS). Procurement now mandates suppliers provide granular emissions factors and verifiable evidence of renewable energy sourcing. 

• Vendor selection heavily prioritizes bids from suppliers with robust ESG governance and formally validated SBTi targets, using contractual mechanisms to enforce supply chain decarbonization and data transparency. 

A framework for sustainable computation in 2026

In 2026, organizations should consider an interdisciplinary program that brings together hardware design, software optimization, energy systems, and economic policy to enable more sustainable computation. Some key steps include:

Reduce embodied carbon via modular, reusable hardware	Design for disassembly & reuse: Require field-replaceable modules, standardized fasteners, and component passports; integrate CXL to reuse legacy DRAM in new pools and reduce fresh memory manufacturing.
	Circular operations at scale: Stand up “circular centers” or partner with OEM asset recovery (Dell, Microsoft) for secure decommissioning, parts harvesting, and resale/remarketing with auditable metrics. Aim for ≥90% reuse/recycle of decommissioned hardware.
	Storage circularity: Institutionalize certified sanitization to achieve ≥80% drive reuse rates; prioritize firmware/health telemetry standards that support predictive refurbish decisions.

Minimize operational carbon through energy-smart compute	Thermal modernization: Deploy liquid cooling where rack densities exceed ~30–40 kW; quantify PUE improvement and WUE impacts; plan for waste-heat recovery to district heating where feasible.
	24/7 carbon-free energy (CFE): Move beyond annual REC matching to hourly matching, coupling PPAs with storage and demand-shifting; locate new capacity in clean-grid regions and cooler climates.
	Workload and network offload: Use schedulers and DPUs/SmartNICs to place inference close to renewable availability and offload data-plane work, cutting CPU cycles and power

Software optimization: do more with fewer joules	AI model efficiency: Favor efficient AI techniques for inference workloads. Track energy use per transaction and set clear thresholds by application. Align this with dynamic right-sizing of infrastructure, using the most suitable systems for each workload to minimize energy consumption and cost.
	Telemetry-driven ops: Implement AI-based cooling control and energy analytics to reduce overhead energy by 20–40% where possible; incorporate grid signals for temporal shifting of non-urgent jobs.

Economic and disclosure policies that reward sustainability	Procurement incentives: Weight tenders ≥30% on sustainability criteria (embodied/operational carbon, circularity, transparency). Require suppliers to maintain SBTi-aligned targets and provide TCFD/ISSB-ready datasets.
	Price signals: Internal carbon pricing for project prioritization; incorporate “emissionality” to favor PPAs that add new clean capacity where grid carbon intensity is highest.
	Compliance readiness: Track legislative developments; even amid U.S. uncertainty, prepare for material Scope 1–2 disclosure and supply-chain (Scope 3) transparency demanded by investors.

Enterprise IT hardware procurement priorities for 2026 

Embed circularity: Mandate take‑back, refurbishment, and component reuse (DRAM/SSDs/PSUs), with auditable reuse ratios and embodied carbon per SKU. Require Digital Product Passports consistent with ESPR. Target ≥90% reuse/recycle of decommissioned equipment.  

Thermal and energy modernization: Specify liquid‑cooling readiness (manifolds, cold plates, leak detection), AI‑optimized cooling control, and PUE/WUE reporting. Favor sites with access to 24/7 CFE pathways and waste‑heat offtake.  

Efficient architectures: Include DPUs/SmartNICs and CXL in baseline requirements; evaluate ARM/x86 mix for perf/Watt and perf/kgCO₂e; assess embodied carbon with tools (SCARIF or vendor LCAs) and account for uncertainty bounds.  

Disclosure and assurance: Require suppliers to provide TCFD‑aligned climate risk disclosures, SBTi‑validated targets, and auditable energy/emissions data (hourly where possible). Prepare for ISSB/CSRD alignment and evolving SEC expectations.   

Risk mitigation: 

• Grid capacity & siting delays: Diversify sites, secure firm PPAs and storage, plan for microgrid options.  

• Thermal transition risks: Manage liquid‑cooling with robust monitoring and maintenance; quantify ROI and leak mitigation.   

• Supply‑chain variance in embodied carbon: Use probabilistic models and STEC adjustments (location/time of manufacture) to avoid misestimation.  

Opportunities: 

• Cost and carbon wins from reuse (components and drives), lower capex via leasing/PaaS with guaranteed return streams, and OPEX reductions through PUE improvements and DPUs offload.   

• Brand and compliance advantage through early alignment with ESPR, ISSB, and SBTi criteria.  

How to prepare: 

Run a 2026 sustainability pre‑qualification across suppliers, including: 

• Circularity scorecards 

• Thermal readiness audits 

• Architecture efficiency benchmarks (perf/Watt, perf/kgCO₂e) 

• Disclosure/assurance maturity – Update category strategies, add internal carbon pricing, and integrate energy aware workload placement into sourcing decisions 

The year ahead for IT hardware procurement  

Sustainable computation in 2026 demands collaboration across computer science, engineering, economics, and environmental science. When procurement links modular design, software efficiency, energy systems, and policy incentives, enterprises can meet rising compute demand and cut carbon – turning sustainability from constraint into competitive advantage. 

References

• European Commission, “Ecodesign for Sustainable Products Regulation (ESPR) 2025–2030 Working Plan,” European Commission, April 16, 2025. Available: EU ESPR Regulation.
• European Commission, “Directive on Repair of Goods (Right-to-Repair Directive),” EUR-Lex, June 2024; updates effective July 2026. Available: EUR-Lex Directive. 

• Microsoft, “2025 Environmental Sustainability Report: Circular Datacenter Hardware,” Microsoft Corporate Sustainability, July 2025. [Online]. Available: Microsoft Circular Centers Report. 

• Dell Technologies, “Streamline IT Asset Refresh for Efficiency: Asset Recovery Services Case Study,” Dell Blog, January 14, 2025. [Online]. Available: Dell Asset Recovery Services. 

• Circular Drive Initiative, “Incentivizing Circular Economy Reuse of Data Storage Drives,” Seagate White Paper, July 2024; updated April 2025 with rare earth recovery pilot. [Online]. Available: Circular Storage Study. 

• International Energy Agency (IEA), “Electricity 2025: Global Data Center Energy Demand Forecast,” IEA Report, February 2025. [Online]. Available: IEA Electricity Report. 

• TrendForce, “Liquid Cooling Adoption in Data Centers: Market Penetration and PUE Gains,” TrendForce Data Center Cooling Report, August 2025. [Online]. Available: Liquid Cooling Best Practices. 

• NVIDIA, “BlueField DPUs for Energy Efficiency in AI Workloads,” NVIDIA Networking Solutions Brief, 2025. [Online]. Available: NVIDIA BlueField Overview. 

• Science Based Targets Initiative (SBTi), “Corporate Near-Term Criteria Version 5.3,” SBTi Standards, September 2025. [Online]. Available: SBTi Criteria Document. 

• ISSB / TCFD, “Climate Disclosure Alignment and Assurance Guidance,” IFRS Sustainability Standards Board, 2025. [Online]. Available: ISSB-TCFD Guidance. 

OxygenIT, “Cloud Carbon Footprint Management in 2025: Compliance & Optimization,” OxygenIT Blog, April 2025. [Online]. Available: Cloud Carbon Footprint Guide.