New Hampshire Data Center Intel

Recent New Hampshire data center news

• Mar 17, 2026

  Jensen Huang Maps the AI Factory Era at NVIDIA GTC 2026

  Nvidia CEO Jensen Huang announced that AI is entering an infrastructure phase and unveiled hardware, software, and reference architectures to build gigawatt-scale “AI factories” for continuous inference.

  • Main announcement: Nvidia unveiled new infrastructure components including Grace Blackwell NVLink72, Vera Rubin (rack-scale systems with ~3.6 exaflops per rack and 45°C hot-water liquid cooling), the Vera Rubin DSX AI Factory reference architecture and Omniverse DSX digital-twin blueprint, and software layers OpenClaw / Nemo / Nemotron to orchestrate and secure agentic AI systems; Nvidia estimated a $1 trillion AI infrastructure market and cited an industry shift to continuous inference.
  • Details & partners: Nvidia described hybrid architectures integrating Groq accelerators (disaggregated inference via Dynamo orchestration), a production co-packaged optical switch built with TSMC, DSX integrations with partners (Cadence, Dassault Systèmes, Schneider Electric, Siemens, Vertiv, Trane Technologies, Switch) and energy partners (GE Vernova, Siemens Energy, Hitachi Energy, Emerald AI); the keynote also referenced venture funding > $150 billion for AI startups and examples like Nestlé reducing compute costs by 83% on a GPU-accelerated workload.
• Mar 13, 2026

  The Gigawatt Bottleneck: Power Constraints Define AI Data Center Growth

  Bloom Energy has released the 2026 Data Center Power Report finding electricity availability has become a defining boundary on data center expansion.

  • Main announcement: The Bloom Energy 2026 Data Center Power Report concludes electricity availability is now a primary constraint for data center growth; it projects U.S. IT load could rise from ~80 GW (2025) to ~150 GW (2028), and highlights major grid forecast revisions such as ERCOT increasing its 2030 data center demand projection from 29 GW to 77 GW and a possible statewide peak of 218 GW by 2031. The report also states roughly one-third of U.S. data centers may rely entirely on onsite power by 2030 and that ~20% of campuses could exceed 1 GW by 2030, rising to nearly 1 in 3 by 2035.
  • Background and details: The analysis is based on surveys of hyperscalers, colocation providers, utilities, and equipment suppliers through 2025 and documents operational shifts: Texas may exceed 40 GW by 2028 (nearly 30% national share); Georgia market share projected +75% while several legacy markets could lose >50% relative share; utilities and developers show a 1–2 year expectation gap on “time to power”; >70% of developers are evaluating onsite power providers; by 2028, 60% expect higher-voltage busways and 45% expect DC architectures.
• Mar 11, 2026

  Meta’s Expanded MTIA Roadmap Signals a New Phase in AI Data Center Architecture

  Meta has outlined and is deploying its MTIA (Meta Training and Inference Accelerator) program, describing roadmaped chip generations and current production use in its data centers.

  • Main announcement/action: Meta outlined a roadmap for successive generations of MTIA processors and reports that first production deployments are already running in its data centers, supporting ranking and recommendation models that generate trillions of predictions per day. Future generations are expected on an ~18–24 month cadence and aim to increase throughput and efficiency for large-scale inference.
  • Background and details: The engineering blog post and industry reporting detail how MTIA lets Meta shape rack-level power and thermal envelopes, integrate with liquid-to-chip cooling, and apply chip-level power management (power capping, workload throttling) to run racks closer to electrical limits; the article also situates MTIA alongside other hyperscaler chips (Google TPUs, AWS Trainium/Inferentia, Microsoft Maia) and notes hyperscale planning for hundreds of megawatts to gigawatt-scale campuses.
• Mar 10, 2026

  From Real Estate to AI Factories: 7x24 Exchange's Michael Siteman on Power, Politics, and the New Logic of Data Center Development

  Data Center Frontier published a podcast episode featuring Michael Siteman (President, Prodigious Proclivities and longtime 7x24 Exchange leader) discussing how AI demand, power scarcity, network strategy, and local politics are reshaping data center development.

  • Main announcement/action: The episode frames site selection as now a systems engineering challenge driven by power availability, network topology, and political risk; Siteman highlights the rapid market shift toward behind-the-meter/onsite generation (usually gas) — noting attitudes changed from “no grid interconnection, no interest” six months ago to willingness to accept onsite generation in the last 30 days. Also cited: rack densities of 120–150 kilowatts and widespread adoption of liquid-to-chip cooling for AI workloads.
  • Background and details: The conversation documents increased pre-leasing of capacity before building completion (with lenders wary because contracts often include termination rights), the labor shortage (“Data centers don’t run themselves”), and a concrete cost datapoint that one high-performance AI server with GPUs and storage can cost over half a million dollars; examples include an island-powered site relying entirely on onsite generation where fiber is ~1,000 feet away.
• Mar 5, 2026

  OpenAI and the Rise of the Multi-Cloud AI Factory

  OpenAI is deepening infrastructure partnerships with AWS to deploy large-scale Trainium capacity and pursuing multi-cloud compute strategies, with reports of very large cloud commitments and potential strategic investments.

  • Main announcement: OpenAI is expanding its infrastructure footprint with AWS, planning to deploy as much as 2 GW of Trainium-based AI compute capacity under an expanded long-term cloud agreement; reporting also suggests the AWS–OpenAI contract could grow by as much as $100 billion in cloud consumption commitments and include a potential $50 billion Amazon investment (reported). These are reported/plausible terms presented as industry reporting rather than a formal OpenAI press release.
  • Background and details: The article synthesizes reporting and analysis: it highlights multi-cloud strategies (Azure + AWS + specialized providers), continued centrality of Nvidia GPUs (and architectures like Vera Rubin), involvement of capital providers such as SoftBank, and mentions other infrastructure partners like Oracle Cloud and CoreWeave; it frames the shift as a transition to gigawatt-scale AI infrastructure and discusses governance questions tied to OpenAI’s nonprofit/for-profit structure.
• Mar 3, 2026

  Powering AI When the Grid Can’t: Inside the New Behind-the-Meter Playbook

  Data Center Frontier hosted a special edition episode of The Data Center Frontier Show (published March 3, 2026) recasting the DCF Trends Summit 2025 session “From Grid to Onsite Powering: Optimizing Energy Behind the Meter for Data Centers.”

  • Main announcement/action: The podcast episode (moderated by Fengrong Li of FTI Consulting) presented a panel discussion with Siemens Energy, Oklo, AlphaStruxure, and ECL that framed behind-the-meter power as critical-path infrastructure for AI deployments rather than contingency capacity; key points include modular phased power blocks (Oklo’s 75-megawatt increments), the need for fast-response buffering technologies (batteries, flywheels, supercapacitors), and the centrality of long-term offtake, capacity reservations, and credit support to unlock equipment and fuel supply.

    • Date: March 3, 2026.
    • Format/location: Podcast episode / online (recap of DCF Trends Summit 2025 session).

  • Background and details: The panel highlighted concrete operational constraints driving onsite power: multi-year utility interconnection delays (“five, ten, even ten-plus years”), urban fuel logistics limits (natural gas pressure, hydrogen delivery scaling), and variability from AI workloads requiring modularity and capital-backed contract structures (capacity reservation deposits, prepayments, offtake commitments) to move projects from planning to execution.

• Mar 2, 2026

  Land and Expand: Early 2026 Megaprojects Reflect a Power-First Ethos

  Data Center Frontier reports multiple developers advancing power-first, land-and-expand AI-ready data center campuses in early 2026.

  • Main announcement/action: Developers including Applied Digital (Delta Forge 1), Vantage (Lighthouse), AVAIO Digital (Little Rock), Rowan (Project Temple), Crow Holdings (Dallas) and Amazon (northwest Louisiana) are advancing large-scale projects that pair land banking with secured power and infrastructure commitments; examples include Applied Digital’s 430 MW Delta Forge 1 (two 150 MW facilities on 500+ acres, first operations targeted 2027) and Vantage’s $15B+ Lighthouse (four hyperscale data centers delivering nearly 902 MW IT load on ~672 acres, construction through 2028).
  • Background and details: Projects feature explicit infrastructure co-investments and timelines: Amazon’s $12 billion Louisiana buildout includes up to $400 million for regional water improvements and 100% developer-funded electric infrastructure; AVAIO’s $6 billion Little Rock hub has a 150 MW Entergy Arkansas commitment with potential to scale toward 1 GW, and Rowan’s Project Temple (300 MW, ~700 acres) targets initial operations in 2027 with ~$700 million local investment and unanimous local approvals.
• Feb 27, 2026 · RMI

  Why Communities Can and Must Consider Electricity Affordability and Risk Together

  Stephen Abbott of RMI argues that communities should consider electricity affordability and risk together and pursue diverse, distributed energy portfolios rather than relying solely on large centralized fossil-fuel generation.

  • Main announcement/action: Communities and local governments should adopt portfolio-based energy strategies (energy efficiency, batteries, renewables, virtual power plants, and other flexible resources) to reduce price volatility and operational risk; RMI highlights concrete examples including data center-driven load growth of 32% by 2030, and Burlington’s 59,204 MWh annual reduction from its energy efficiency program.
  • Background and details: The piece cites recent cost and risk evidence: ComEd provided $277 million (2024) for efficiency programs yielding an estimated $3.2 billion in customer savings; reliance on fossil fuels produced at least $390 million in excess costs for communities around the Prairie State Energy Campus over four years; typical monthly fuel charges in Florida doubled from ~$20 to ~$40 (2020–2023); utilities such as TVA are proposing large new gas facilities as a conventional response.
• Feb 27, 2026

  Policy Shock: Big Tech Told to Power Its Own AI Buildout

  The White House is advancing a ‘ratepayer protection’ framework aimed at ensuring large AI data center projects do not shift grid upgrade costs onto residential customers.

  • Main action: The White House is pushing a ratepayer protection approach that would encourage/require large AI and hyperscale developers to demonstrate energy self-sufficiency or provide dedicated power solutions (e.g., behind-the-meter generation) when seeking large-load approvals; the article cites signals that formal guidance or rulemaking and possible state-level measures could follow in the near term.
  • Context and details: The article reports market movement (about one-third of new U.S. projects evaluating private/on-site power), technical choices include natural gas turbines, fuel cells, hybrid microgrids, and renewables, capacity scales of hundreds of megawatts to gigawatt levels are discussed, and a cited Nordic deal (Equinix/atNorth) reports roughly 1 gigawatt of secured power capacity and further expansion plans; potential near-term indicators include utility tariff changes, hyperscaler commitments, and federal guidance.
• Feb 26, 2026

  Cooling’s New Reality: It’s Not Air vs. Liquid Anymore. It’s Architecture.

  Data Center Frontier (Matt Vincent) summarizes a batch of industry announcements and product launches that collectively reframe data center cooling as a full-stack systems engineering challenge.

  • Summary of main announcements: HRL Laboratories unveiled Low-Chill (Feb. 24, 2026), a single-phase direct liquid cooling approach developed under DOE/ARPA-E that claims +40% processor cooling or >10X reduction in pumping power, while Johnson Controls agreed to acquire Alloy Enterprises (Feb. 18, 2026) (expected to close in fiscal Q3; financial terms undisclosed). Carrier (Feb. 26, 2026) and Modine/Airedale (Jan. 22, 2026) launched chillers emphasizing –20°F to 140°F operating range, fast recovery, and hybrid free-cooling; Infinium (Jan. 15, 2026) launched Infinium Edge immersion platform; Boyd announced a manufacturing expansion in Juarez to ~460,000 sq ft (Feb. 17, 2026); Waste2Nano announced a wastewater-cooled AI platform targeting 10,000–20,000 m³/day (~5 MGD) initial deployment.
  • Background and supporting details: The article is a roundup/opinion-style synthesis (not a single primary press release) that compiles multiple company announcements and trade-show reveals from Jan–Feb 2026, highlights thermal metrics disclosed by HRL (e.g., 8.2 °C/kW thermal interface resistance; <1 psi pressure drop; <1% pumping power block-level; 70°C inlet) and firm product claims (Johnson Controls: up to 35% thermal efficiency improvement, up to 75% pressure-drop reduction). It notes regulatory/transaction timing (JCI/Alloy closing subject to regulatory approvals in fiscal Q3) and clarifies which items are product launches versus strategic acquisitions or manufacturing expansions.