Getting your news
Attempting to reconnect
Finding the latest in Climate
Hang in there while we load your news feed
California Data Center Intel
Latest data center news, projects, power and policy across California — updated daily.
California · Construction & power moves · 1
full tracker →Land, power, and interconnection moves across California — each traced to primary filings.
Top JUST IN — California
-
CAISO schedules Large Loads Initiative stakeholder meeting and draft final proposal comment period
Source: California ISO · Sep 10, 2026According to CAISO, the ISO will host a hybrid stakeholder meeting on Oct. 1, 2026 for the Large Loads Initiative, post a draft final proposal on Sep. 24, 2026, and open comments due by Oct. 15, 2026. CAISO also says stakeholders will review the proposal before it goes to the Board of Governors and before the ISO files tariff language with FERC, indicating an active regulatory process for large loads in California.
Backed by 1 primary filing — sign in or book a call to see all sources.
-
Large Loads Initiative: Additional discussion of operational and technical requirements, call on 8/24/26
Source: California ISO · Aug 17, 2026California ISO said it scheduled an additional stakeholder meeting for the Large Loads Initiative “to provide additional time, if needed, to discuss the operational and technical requirements” in the Large Loads Straw Proposal and Large Loads Technical Requirements Straw Proposal, with topics including “Operational requirements” and “Technical requirements” (California ISO, Aug. 17, 2026). The notice also says the Aug. 24 meeting was contingent and ultimately unnecessary because “We were able to cover all of the material and will no longer need the meeting on 8/24.”
Backed by 1 primary filing — sign in or book a call to see all sources.
-
Large Loads Initiative: Straw proposal posted, meeting on 8/19/26
Source: California ISO · Aug 12, 2026California ISO says that, following the FERC “June 18, 2026, Order to Show Cause,” its Large Loads Initiative is the “primary forum” for possible “tariff, policy, and process enhancements” affecting large-load integration on the ISO-controlled grid, including “planning, interconnection, transmission service, reliability, and operational considerations.” CAISO also says the straw proposal will address “Cost Shifting Risk Among Transmission Customers,” “Co-Location Arrangements and Load with Behind the Meter Generation,” and “Interconnection Customers Serving Electrically Proximate Large Load and Co-Located Load.”
Backed by 1 primary filing — sign in or book a call to see all sources.
-
Large Loads Considerations: Comment deadline on information session moved to 2/25/26
Source: California ISO · Feb 12, 2026California ISO said it “has moved the deadline for written comments on the Large Load Considerations issue paper and Feb. 5 meeting discussion to Feb. 25, 2026,” and asked stakeholders to submit comments by end of day Feb. 25, 2026. The notice also says the “issue paper, presentation, and meeting recording are available” on CAISO’s Large Load webpage.
Backed by 1 primary filing — sign in or book a call to see all sources.
-
2025 Draft Participating Transmission Owner Per Unit Cost Guides Posted
Source: California ISO · Apr 03, 2025CAISO said it “has posted the Draft Per Unit Cost Guides” and that “participating transmission owners update and publish per unit costs annually as part of the current interconnection study cycle for estimating the cost of facilities required to interconnect resources,” with written comments due by April 30, 2025, according to the California ISO notice.
Backed by 1 primary filing — sign in or book a call to see all sources.
Recent California data center news
-
Nvidia unveils alternative high bandwidth technology to bolster AI cards
Nvidia has announced NVHBM, a customized High Bandwidth Memory architecture for AI accelerators and custom XPUs.
- The company says NVHBM is part of its NVLink Fusion platform and is aimed at hyperscalers and AI companies building custom accelerators; Amazon’s Annapurna Labs is named as the first announced partner.
- Nvidia says the design moves the memory controller into the HBM stack’s base die, and claims up to 30% greater bandwidth, 15% lower power, and 25% more compute-die area versus conventional HBM4E.
- The article frames this as a blog-post announcement published alongside Nvidia’s quarterly earnings call, not as a report of a separate commercial deal.
-
Nokia to Transform NXP’s Chandler Campus for AI-Era Optical Chips
Nokia has announced a definitive agreement to acquire NXP Semiconductors’ Chandler Fab and Chandler Probe site and repurpose it for AI infrastructure-related optical component manufacturing.
- Nokia will lease part of the facility in early 2027 and acquire the full campus in Q1 2029, subject to regulatory approval.
- The site will produce indium phosphide semiconductors for optical components used in high-speed networks linking processors, memory, and systems inside AI data centers.
- Nokia said its Optical Networks sales increased 20% YoY in Q2 and sales to AI and cloud customers increased 105%.
- NXP issued a discontinuation notice for legacy products made at Chandler that will not move to another manufacturing site, and said it does not expect near-term supply impact during the transition.
- Nokia is also preparing to ramp a new San Jose fab and expand advanced testing and packaging capacity in Pennsylvania tenfold.
-
Solid-State Transformers Power Next-Gen AI Data Centers
The article explains the emerging role of solid-state transformers in high-voltage direct current power architectures for AI data centers, while citing industry views and ongoing pilots rather than announcing a single new commercial deal.
- Delta says its SST platform converts MV AC (6.6-35 kV) directly to 800 V DC at up to 98.5% efficiency; it was developed with Open Compute Project efforts and Nvidia’s 800 V DC specification and was demoed at the OCP Summit 2025.
- Alderbuck Energy is piloting its technology at the San Diego Supercomputer Center at UC San Diego; Rick Sander said he expects most new installations to use SSTs by 2028 and most data centers to rely exclusively on SSTs by 2035. The article also presents cautionary views from SPOC Energy and Hitachi Energy on maturity and supply-chain readiness.
-
United States and Cameroon Announce $7 Billion in Investment and Trade Opportunities at Inaugural Economic and Commercial Dialogue
The United States and Cameroon have announced approximately $7 billion in new and expanding investment and trade opportunities during the inaugural U.S.-Cameroon Bilateral Economic and Commercial Dialogue.
- The announcement covers sectors including technology, critical minerals, infrastructure, energy, forestry, and logistics, and the two sides agreed to establish joint working groups and continue the Dialogue ahead of the next meeting in 2027, which the United States will host.
- Specific projects mentioned include Tronox/Lion Rock‘s $500 million rutile project, ARM‘s planned $85 million cobalt-nickel-manganese project in Nkamouna, Hydromine‘s proposed $3.5 billion, 1,040-MW Grand Eweng Hydropower Project, and Cybastion‘s $75 million AI-powered data center and power plant in Douala.
-
AI Rack Density’s Real Limits: Power, Cooling, Failure Risk
The article analyzes how AI is pushing data center rack density higher and argues that cooling, power delivery, and grid access now set the main limits on AI rack design.
- It cites Uptime Institute data showing modal rack density rose to 11 kW in 2026 from 9 kW in 2025, while Nvidia GB300 NVL72 reaches up to 142 kW per rack and Rubin Ultra NVL576 “Kyber” is specified at roughly 600 kW for H2 2027.
- It says direct-to-chip liquid cooling now handles 100-150 kW per rack and is the dominant liquid method; 800 VDC is shipping with Vera Rubin NVL72, and vendors including Vertiv, Schneider Electric, Eaton, and Delta have commercial 800 VDC offerings slated for H2 2026. It also references Foxconn’s 40 MW Kaohsiung-1 facility in Taiwan, the Open Compute Project Mount Diablo specification finalized in March 2026, and Microsoft/Meta hardware demonstrations in July 2026.
-
Pass the Salt: A Sodium-Iron Battery Breakthrough
Inlyte Energy discussed its sodium-iron battery technology and announced the shipping of its first battery pack to Southern Company’s Wilsonville, Alabama energy storage facility.
- The company said it is shipping its first utility-scale pilot: a 1-megawatt-hour battery for Southern Company, with commissioning expected in the next couple months at the Wilsonville, Alabama test site.
- Antonio Baclig said Inlyte acquired Beta Research in 2022, produces cells in the UK near Derby, and is using iron powder and food-grade table salt to enable safer, lower-cost batteries for grid-scale and data center use.
- Baclig added the system can provide 24 to 48 to 72-hour backup for data centers, may fit a 3-megawatt, 24-hour duration backup on a diesel pad, and benefits from U.S. battery tax incentives supporting domestic manufacturing.
-
The Janus Five: 20+ Commercial Nuclear Microreactors, Starting in 2028
The U.S. Army has announced the Janus Program selections for more than 20 commercial nuclear microreactors across five military installations.
- The Army selected BWXT Advanced Technologies for Fort Campbell, Kentucky; Radiant Industries for Fort Benning, Georgia; General Atomics Electromagnetic Systems for Fort Hood, Texas; Antares Nuclear for Fort Bragg, North Carolina; and Westinghouse Government Services for Fort Drum, New York.
- The five vendors may draw on up to $2.2 billion in Army investment through fiscal year 2031 under fixed-price, milestone-based OTA agreements negotiated through DIU; first project service is targeted for September 2028.
- The article is a news report/commentary summarizing a first-time announcement and includes background on related Pentagon microreactor programs, including the Air Force’s ANPI and Project Pele.
- Several vendor-specific details are provided, including Radiant’s up to $750 million award for 15 Kaleidos units, and Antares’ combined book of around $1 billion across programs.
-
How Data Centers Are Using AI to Run Cooler and Smarter
The article explains how data center operators are using AI to improve cooling, power use, maintenance, and incident response, and how the industry is moving from pilots toward broader production use.
- Digital Realty uses its OPDaaS platform and other AI tools to analyze sub-second telemetry from power, cooling, and building systems; it is also deploying Phaidra at IAD51 in Northern Virginia and plans to bring about 30 more data centers online by the end of 2026.
- AWS, DPR Construction, and Schneider Electric are also applying AI for network operations, construction planning, robotics, digital twins, DCIM, and service dispatch; analysts say most operators still keep humans in the loop and full autonomy is not yet common.
-
US Army to spend $2B to build nuclear microreactors at 5 bases as US seeks to ramp up nuclear power
The U.S. Army has announced plans to add nuclear microreactors at five military bases under its Janus program, selecting five companies to own, construct and operate the reactors if milestones are met.
- The selected sites are Fort Bragg (North Carolina), Fort Campbell (Kentucky), Fort Hood (Texas), Fort Benning (Georgia), and Fort Drum (New York); each reactor is expected to produce 1 MW to 20 MW.
- The Army said the companies could receive up to $2.2 billion total over five years; Radiant said its award is $750 million. The Army also said it is working with the Energy Department on radioactive waste removal and expects billions of dollars in private capital investment.
-
Intel unwraps three-pronged architecture strategy to go after agentic AI
Intel has outlined a broad hardware strategy for agentic AI at Hot Chips 2026, positioning CPUs, GPUs, memory, interconnects and packaging as a combined system approach rather than a GPU-only model.
- Diamond Rapids is Intel’s next-generation Xeon Scalable 7 CPU for enterprise-scale agentic AI, built on 18A-P with up to 256 CPU cores, 1.28GB last-level cache, 16 memory channels at up to 12,800 MT/s, and 128 lanes of PCIe 6.0 and CXL 3.0.
- Crescent Island is a low-power, air-cooled data center GPU for inference with 32 Xe3P-based Xe cores, 256 XMX engines, up to 480GB LPDDR5X, and a 350W TDP; Wildcat Lake targets Core Series 3 processors with 18A, integrated Xe3 graphics, XMX AI acceleration, and an NPU up to 17 TOPS.