Intel, SambaNova Bet on Split Inference as Agentic AI Strains GPUs
Intel and SambaNova announced a heterogeneous inference architecture designed for agentic AI workloads, dividing tasks among GPUs for prompt prefill, SambaNova’s reconfigurable dataflow units (RDUs) for token decode, and Intel Xeon 6 CPUs for orchestration. The three‑tier model aims to alleviate rising inference costs, power limits, and under‑utilized GPU capacity that enterprises face when deploying multi‑step coding agents and other interactive AI systems. Availability is slated for the second half of 2026 and targets cloud providers, sovereign AI deployments, and large enterprises. The partnership emphasizes efficiency over raw performance, seeking higher asset utilization across the data‑center stack.
Intel Joins Musk’s Terafab as AI Compute Race Expands to Space
Intel announced its participation in Elon Musk’s Terafab initiative, a joint effort with Tesla, SpaceX and xAI to build massive AI chip capacity. The project targets producing one terawatt of compute per year, leveraging Intel’s design, fabrication and advanced packaging...
Nutanix Targets Neocloud AI Economics With Platform Push
Nutanix announced a major upgrade to its Nutanix Cloud Platform, branding it for the "agentic AI era" where autonomous workloads generate unpredictable demand. The rollout adds bare‑metal Kubernetes (NKP Metal), multi‑tenant AI tooling via SP Central, and deeper integrations with NetApp, Cisco...
Cold-Climate Data Centers: The Next Hot Thing in Data Center Growth
Cold‑climate data centers exploit naturally low ambient temperatures to provide free cooling, dramatically lowering power and water consumption. Operators such as atNorth, Facebook, and Google have already deployed facilities in Iceland, Sweden, and Finland, demonstrating cost and sustainability gains. The...
IBM, Arm Target Enterprise AI With Mixed-Architecture Approach
IBM and Arm announced a partnership to run Arm‑native applications on IBM Z mainframes and LinuxOne servers through a shared software layer and virtualization. The solution lets enterprises deploy AI workloads across both architectures without rewriting code, preserving the uptime,...
Neocloud Storm Gathers as Data Center Deals Stall Over Credit Risk
Neocloud providers are finding that aggressive pricing, long‑term terms and prepaid fees no longer secure colocation capacity. Providers now demand investment‑grade credit and solid utilization visibility, turning lease underwriting into a project‑finance exercise. Even premium rates of $155‑$160 per kW...
Is Multi-Line Insurance a Better Fit for Data Centers?
Data centers confront property, cyber and liability risks, prompting insurers to offer multi‑line bundles that combine these exposures into a single contract. Recent launches by Aon, FM Global and ATA illustrate growing capacity for such packages. Multi‑line policies can lower...
Knocking on Quantum’s Door: QuiX Claims Photonic Error Reduction Breakthrough
QuiX Quantum announced the first below‑threshold error mitigation on a photonic quantum computer, using a 20‑mode processor and a photon‑distillation gate. The technique achieved a 2.2× reduction in photon‑indistinguishability error and a net 1.2× overall system‑error decrease. Collaborators include NASA’s...
New Data Center Developments: April 2026
Data center capital spending is surging worldwide, with the six largest U.S. hyperscalers projected to invest about $700 billion in 2026, a six‑fold increase from 2022. Major projects include Meta’s $10 billion, 1 GW El Paso facility slated for 2028, a 900 MW AI center...
Gridlocked: How Power Constraints Are Shaping the Future of Data Centers
Power availability, not land or capital, is now the primary constraint on U.S. data‑center expansion. The Department of Energy estimates that half of the 100 GW of new peak capacity needed by 2030 will be driven by data centers, while developers...
Leadership Updates: Key Data Center & Cloud Appointments (Q2 2026)
The data center and cloud sectors are undergoing a wave of senior‑level appointments as operators scale to meet surging hyperscale, AI and sustainability demands. Stream Data Centers promoted Michael Lahoud to CEO, while Prime Data hired power veteran John Bates...
North American Data Center Growth Shifts Toward Execution, Not Expansion
The North American data center market is moving from pure scale‑driven expansion to a phase where execution risk and delivery pathways dominate. Metro Atlanta has emerged as the fastest‑growing U.S. market, while Virginia’s once‑dominant hub is fragmenting under regulatory and...
Inside the Push to Bring DC Power to Data Centers
The Current/OS Foundation and the Open Direct Current Alliance have signed an MoU to coordinate technical work on DC power distribution for data centers. Their collaboration aims to create unified standards, leveraging IEC circuit‑breaker rules and upcoming NFPA code updates,...
Is Microfluidics Ready to Cool the Next Generation of Data Centers?
Data centers face localized hot spots that conventional air cooling cannot efficiently address, prompting interest in microfluidic cooling that routes liquid through microscopic channels etched into silicon. Demonstrations show temperature reductions of up to 80% compared with air, promising higher...
Rebellions Raises $400M to Scale AI Inference, Targets US Expansion
South‑Korea‑based AI hardware startup Rebellions announced a $400 million pre‑IPO round, bringing its total funding to $850 million and valuing the company at roughly $2.34 billion. The financing underlines a market pivot from model training to AI inference, where power efficiency and cost...
Veritone Leans Into Oracle Cloud to Scale AI Data Pipelines
Veritone announced a multi‑year agreement to migrate its core AI workloads, including aiWARE, Data Refinery, and Data Marketplace, to Oracle Cloud Infrastructure. The move aims to boost performance, security, and global scalability as the company tackles massive unstructured data volumes....
Europe’s Digital Networks Act: A Foundation for the AI-Driven Economy
The European Commission introduced the Digital Networks Act (DNA) in January 2026 to replace the 2018 Electronic Communications Code and modernize EU telecom rules. The legislation targets accelerated deployment of 5G/6G, fiber, and secure, resilient infrastructure to meet soaring AI‑driven...
Equinix Targets Talent Gap as AI Infrastructure Demand Surges
Equinix announced a global expansion of its workforce‑development initiatives to address the growing talent shortage in data‑center operations as AI demand accelerates. The company is scaling its Pathways to Tech program, launching a Data Center Technician Training Coalition in Brazil,...
Vertiv Targets AI Cooling Bottleneck with ThermoKey Deal
Vertiv announced it will acquire Italy‑based heat‑exchanger specialist ThermoKey, extending its reach across the entire thermal chain of AI‑intensive data centers. The deal targets the emerging bottleneck of facility‑level heat rejection as power densities rise, shifting focus from chip‑level cooling...
Australia Puts AI Data Centers on Notice With New Approval Rules
Australia has introduced a national framework that ties AI data‑center approvals to renewable‑energy investment, grid upgrades, water stewardship and local economic impact. The policy shifts from a market‑driven, speed‑focused model to one where compliance with national priorities determines access to...
How Stadium Data Centers Power Fans, Operations, and Broadcast
Stadiums hosting the 2026 FIFA World Cup are deploying dual data‑center architectures—one for everyday IT services and another for high‑bandwidth media production. Cisco’s IP Fabric for Media and HPE’s Mist AI platform enable ultra‑low‑latency 100 Gbps video streams and AI‑driven network...
Super Micro Indictment Highlights AI Infrastructure Supply Chain Risks
A federal indictment alleges Super Micro executives facilitated moving Nvidia AI chips to China, prompting co‑founder Yih‑Shyan Liaw's resignation. The case highlights the clash between soaring AI hardware demand, tightening export controls, and supply‑chain integrity. Analysts warn that reliance on...
Nokia Sets Early Wi-Fi 9 Vision Around AI, Real-Time Connectivity Demands
Nokia has announced an early vision for Wi‑Fi 9 that shifts the focus from peak‑throughput to deterministic, low‑latency performance required by AI‑driven and immersive applications. The company targets sub‑10 ms bounded latency, multi‑gigabit delivery under load, and energy‑per‑bit efficiency, positioning Wi‑Fi 9 as...
A Buyer’s Guide to Data Center Colocation Space
The guide breaks down colocation fundamentals, from renting rack space and power to the responsibilities of hardware setup and management. It explains pricing models—monthly, pay‑as‑you‑go, or long‑term contracts—and how space is measured in rack units, with half‑ or full‑rack options...
Ampere Expands European Cloud Footprint Amid Sovereign Cloud Demand
Ampere Computing is rolling out its Arm‑based AmpereOne and AmpereOne M processors across Europe, adding new cloud instances on hyperscale provider Oracle and regional players such as Scaleway, Glesys, CloudSigma, C41.ch and Hetzner. The expansion targets AI inference workloads, power‑limited environments...
Germany’s Data Center Market Expands Amid Power, Regulatory Pressures
Germany’s data‑center market is entering a rapid expansion phase as AI, 5G and cloud workloads drive demand across the country. Frankfurt remains the dominant hub with about 745 MW of live IT load and 542 MW under construction, while Berlin is emerging...
Hyperscaler Capex Snowballs Toward $700B as Firms Stage AI Builds
Moody’s forecasts that the six largest U.S. hyperscalers will spend about $700 billion on capital expenditures in 2026, roughly six times the 2022 level. The spending surge is driven by exploding demand for AI compute capacity to train models and support...
Meta, Nebius Sign $27B Deal to Power Nvidia Vera Rubin Deployments
Meta and Nebius Group have inked a five‑year infrastructure supply agreement valued at up to $27 billion. Nebius will deliver roughly $12 billion of dedicated AI compute built on Nvidia’s upcoming Vera Rubin platform, with new clusters slated for early 2027. Meta...
Preparing Enterprise Data Centers for AI Adoption
A 2025 McKinsey forecast puts AI‑related IT infrastructure spending at nearly $7 trillion through 2030, with $3 trillion earmarked for data‑center capacity. Enterprises must juggle this surge with traditional workloads, requiring distinct strategies for power‑intensive AI training and latency‑critical AI inference. Training...
CoreWeave’s Flexible AI Cloud Pricing Model Signals Strategic Shift
CoreWeave unveiled a flexible AI‑cloud pricing suite that adds “Flex Reservations” and “Spot” instances to its existing reserved and on‑demand options. The new tiers let customers pay a reduced holding fee for guaranteed capacity ceilings and use‑based rates only when...
Building the Foundations of AI Cities: Lessons in Infrastructure, Energy, and Efficiency
AI’s exponential compute needs are outpacing traditional data‑center infrastructure, especially in legacy hubs like Northern Virginia. Operators are now targeting locations where clean power, talent, and renewable resources converge, birthing “AI cities” that embed digital infrastructure into urban planning. Rio...
RDHx: A Simple, Affordable Path to Lower Data Center Energy Use
Rear‑door heat exchangers (RDHx) replace traditional rack doors with liquid‑cooled coils, capturing server exhaust at the source. By moving heat to a closed‑loop liquid circuit, RDHx reduces reliance on room‑level mechanical cooling and can enable higher supply‑air temperatures. The technology...
How Data Centers Rewrite the Playbook for Building Protection
Data center construction is shifting toward vertical, multi‑story facilities, driving stricter fire‑protection, corrosion control, and flooring requirements. Builders are increasingly using modular, factory‑applied protective coatings to ensure consistent performance and accelerate schedules, especially in rural sites with harsh environmental exposure....
AI ‘Virtual Residents’ Offer Early Read on Data Center Sentiment
University of California, Riverside researchers deployed large‑language‑model agents as “virtual residents” to capture community sentiment on data‑center proposals early in the planning cycle. The AI agents consistently highlighted water consumption, utility costs and electricity demand as primary concerns, while identifying...
Retrofits Vs. Rebuilds: Approaches to Adapting Legacy Data Centers for AI
The surge in AI workloads is prompting data‑center operators to consider retrofitting legacy facilities rather than constructing purpose‑built AI campuses. Retrofitting can accelerate deployment, lower capital outlay, and improve sustainability, but it hinges on upgrading power delivery, cooling capacity, rack...
Retrofits Vs. Rebuilds: Approaches to Adapting Legacy Data Centers for AI
The AI boom is driving demand for massive compute capacity, prompting operators to consider both new purpose‑built AI data centers and upgrades to existing facilities. Retrofitting legacy data centers can be faster, cheaper, and more sustainable, but only if power,...
How Data Centers Are Adapting to Extreme Weather
U.S. data‑center operators are confronting a surge in extreme weather events, from floods and wildfires to record heat and cold spells. Scenario analyses from S&P Global and the World Economic Forum estimate climate‑driven costs could equal 9.5% of total data‑center...
Urban Vs. Rural: Why Data Centers Are Built Where They Are
Data center development in the United States is moving beyond the traditional urban corridors of Northern Virginia, Silicon Valley, and Chicago. Expanding power capacity, new long‑haul fiber routes, and aggressive state incentives are making rural states such as Pennsylvania, Louisiana,...
UAE Data Centers: Powering the Middle East’s AI and Cloud Revolution
The United Arab Emirates is rapidly emerging as a pivotal data‑center hub for the Middle East and Africa, with live capacity surpassing 376 MW in 2025. Hyperscale players such as Microsoft, G42, and OpenAI are expanding AI‑focused facilities, targeting an additional...
Google Teams Up with CTC Global for Grid Intelligence
Google Cloud and Alphabet’s moonshot project Tapestry have deepened their partnership with CTC Global to launch GridVista, an observability platform that embeds optical‑fiber sensors in transmission conductors. The system delivers real‑time strain, temperature and vibration data, feeding it into Google...
Edge Data Centers Vs. Edge Devices: When to Use Each
Edge computing can be delivered via purpose‑built edge data centers or through distributed edge devices such as gateways, sensors, and consumer hardware. Data centers provide consolidated compute, storage, and robust security for high‑throughput, latency‑sensitive workloads, while devices excel at mobile,...
The Myth of ‘Always On’: Confronting Data Center SPOFs
Recent incidents—a 2021 Texas freeze and a 2021 OVH fire in Strasbourg—highlight hidden single points of failure in data centers. The freeze delayed fuel deliveries, exposing over‑reliance on on‑site fuel, while the fire demonstrated how passive cooling designs can unintentionally...
AWS Networking Boss Talks Roadmap, Hollow Core Fiber, and Data Center Future
Amazon Web Services announced a $200 billion capital‑expenditure plan through 2026, with networking as a primary focus to support its AI push. The company is trialing hollow‑core fiber in a handful of locations to extend latency‑critical inter‑zone links, while a newly...
Voltage Ride-Through: A Key Ingredient in Data Center Resilience
Voltage ride‑through (VRT) enables data‑center hardware to stay connected to the grid during brief outages or brownouts, preventing immediate shutdowns. By using UPS units that buffer power and can switch instantly to backup sources, VRT eliminates the latency of manual...
AI Infrastructure Financing Enters a New Era: What Execs Need to Know
AI infrastructure financing is rapidly evolving as traditional bank lending can’t keep pace with the capital intensity of AI‑ready data centers. Lenders are stepping back because power‑delivery risk and SPV structures clash with cash‑flow lending models. GPUs now face six‑to‑nine‑month...
Passing the Torch: Building a Workforce for the Next Generation of Data Centers
The data‑center industry faces a looming talent shortage as up to half of its engineers could retire within the next two years, creating a critical experience gap. Rapid growth in AI‑driven workloads and ultra‑high‑density designs intensifies the need for skilled...
Forget Quantum? Why Photonic Data Centers Could Arrive First
Photonic computing is emerging as a realistic path to higher‑throughput, more energy‑efficient data centers, potentially arriving before general‑purpose quantum machines. By using photonic integrated circuits to perform linear‑algebra operations in the optical domain, these systems promise faster speeds, greater bandwidth,...
The Data Center Surge Has a Hidden Source of Carbon Emissions
Data center construction will need 2 million metric tons of cement by 2030, potentially releasing 1.9 million tons of CO₂ if conventional concrete is used. Tech giants such as Microsoft, Amazon and Meta have signed low‑carbon concrete offtake agreements with startups like...
AmberSemi Launches PowerTile to Cut Data Center Power Drain
California fabless chipmaker AmberSemi announced its new PowerTile, a quarter‑size, 1,000‑amp vertical power‑delivery module designed to sit behind AI processors in servers. The device claims to cut board‑level power distribution losses by 85%, potentially saving 225 MW of electricity per year...
Why Your AI Chip Utilization Problem Is Really a Storage Problem
AI performance hinges not just on GPUs or LLMs but on the storage layer that feeds data to accelerators. A Meta‑Stanford white paper shows storage can consume up to one‑third of the power used for deep‑learning training. When storage cannot...
