Indian cloud provider NxtGen AI adopts diamond-cooled GPU servers

An Indian cloud provider has adopted Nvidia GPU servers cooled using synthetic diamonds.

Akash Systems this week announced it has delivered the world's first diamond-cooled GPUs to NxtGen AI, an Indian cloud provider.

NxtGen has received Nvidia H200 GPU servers equipped with Akash's proprietary diamond cooling technology.

Full details on the number of servers and where they have been deployed haven't been shared.

California-based Akash says its cooling tech is projected to reduce data center energy consumption and increase FLOPs/Watt by up to 15 percent per server in facilities running at temperatures up to 50°C (122°F), compared to the common standard of 24-29°C (75-85°F).

Diamond is a very efficient thermally conductive material, enabling heat to be removed 5x faster than the current industry standard, copper. Akash uses synthetic diamond materials on the chips to remove heat more efficiently than copper. The company claims to be able to reduce GPU hotspot temperatures by 10°C (50°F).

NxtGen offers a number of IT services, including cloud, CDN, colocation, and disaster recovery. The company reportedly currently operates data centers across Ahmedabad, Bangalore, Faridabad, and Mumbai

Reports last year said the two companies had signed a $27 million deal for the supply of the diamond-cooled GPUs.

Akash is backed by venture capital investors Khosla Ventures and Founders Fund. The firm has secured more than $20 million across six funding rounds.

Nexalus picks Alps Alpine to manufacture liquid cooling kit for data centers

Liquid cooling vendor Nexalus has selected electronics firm Alps Alpine as manufacturing and systems integration partner for its data center thermal management products.

The partnership will marry Nexalus’s advanced science, thermal engineering, and system-level design capabilities with Alps Alpine’s automotive-grade manufacturing, quality systems, and global operational footprint.

Together, the companies “form a unified platform capable of delivering integrated Nexalus thermal management systems consistently and reliably anywhere in the world,” according to a joint statement announcing the deal.

Alps Alpine will serve as manufacturer and integrator of complete Nexalus systems, supporting advanced thermal management assemblies, integrated ICT infrastructure solutions, and full platform builds.

Kenneth O’Mahony, CEO of Ireland-based Nexalus, said: “Alps Alpine’s world-class manufacturing, quality systems, and global operational footprint provide the execution backbone required to deliver our custom-designed platforms at global scale. Their approach to reliability, traceability, and performance aligns directly with our global customers’ mission-critical demands of next-generation digital infrastructure.”

Cooling equipment will be manufactured at Alps Alpine’s Millstreet facility in County Cork, Ireland, originally established as an ICT manufacturing hub for Apple.

Nexalus offers direct-to-chip liquid cooling solutions, and says its systems can recycle excess heat so it can be used in applications such as district heating.

Prometheus Hyperscale details plans for Dallas data center campus

Wyoming data center firm Prometheus Hyperscale is targeting a new development in Texas.

Neilson’s post suggests the liquid-cooled site could reach 200MW of utility load and feature four large buildings.

The company is set to use behind-the-meter power through Jenbacher's J620 gas engines and will be connected to Engie’s adjacent battery energy storage system.

Launched around 2020 and formally known as Wyoming Hyperscale Whitebox, Prometheus was relaunched in September 2024. The Wyoming-based company initially planned to develop a 120MW campus back in 2020. However, after its relaunch, plans were revised to 1GW with plans to target multiple sites, including Arizona and Colorado.

In September 2025, Prometheus partnered with Engie to colocate data centers with renewable assets in Dallas, Texas, though full details weren’t shared.

Backed by private equity firm Grey Rock Investment Partners, Conduit Power provides on-site natural gas and battery storage assets. Conduit first announced a partnership with Prometheus in September 2025 to deploy power solutions in Texas.

In November, the firm announced purchase orders for 75 Ecomax 33 power generation units, powered by Jenbacher 620 engines, with AB Energy.

Johnson Controls to acquire Alloy Enterprises

Johnson Controls has bought a manufacturer of cold plates and other liquid-cooling components.

The company this week announced it has signed an agreement to acquire Alloy Enterprises, a Boston-based company specializing in thermal management components for data centers and other mission-critical industrial applications.

Terms were not shared. The deal is expected to close in Q3 2026.

Founded in 2020, Alloy Enterprises is a thermal, mechanical, and materials sciences technology firm. Alloy fabricates aluminum and copper components using its patented Stack Forging manufacturing process, creating cold plates, heat pipes, and other components. Johnson said the move will expand its capabilities in the growing data center cooling segment.

Johnson Controls provides and deploys data center solutions across more than 150 countries. In 2024, it launched a dedicated data center solutions business to provide integrated solutions to data center customers on a global scale. It offers many data center cooling products through its Silent-Aire unit.

Johnson has previously invested in two-phase direct-to-chip liquid cooling firm, Accelsius.

Macquarie Technology upsizes debt capacity by AU$50 million to expand IC3 Super West facility in Sydney, Australia

Macquarie Technology Group has completed a AU$50 million (US$35.3 million) incremental debt facility, increasing its secured revolving loan facility to US$353 million, to expand its flagship data center development in Sydney, Australia.

The funds will be used to expedite capacity improvements for the IC3 Super West facility in Macquarie Park, facilitating the acquisition of long lead-time equipment and bringing the site from its initial 6MW of capacity to 19MW. The facility has a total planned capacity of 47MW.

Macquarie said this approach “aligns with current market dynamics, which is seeing customer demand for larger data center capacity and faster lead times.”

The debt facility is arranged under the existing agreement.

Macquarie subsidiary Macquarie Data Centres broke ground on IC3 Super West in June 2024, with plans for the facility to support both air cooling and direct-to-chip liquid cooling.

It is the third and final data center on the Macquarie Park data center campus. Phase I of the build delivered the complete core and shell with a 6MW load for an investment of US$247 million.

The facility is due to open in September 2026.

OpenAI partners with Tata to deploy data center capacity in India

The Tata Group, Tata Consultancy Services (TCS), and OpenAI this week announced a new strategic partnership.

TCS’ HyperVault unit and OpenAI have agreed to a multi-year partnership to develop AI infrastructure in India. In the initial phase, TCS will develop AI infrastructure with 100MW of capacity, with an option to scale to 1GW.

HyperVault was announced during TCS’ Q2 2025 results in October, with plans to develop 1GW of liquid-cooled data center capacity across India, though few details have been shared. Private equity firm TPG has invested in the venture.

The agreement will also see Tata employees gain access to Enterprise ChatGPT, and the two companies will co-develop industry-specific AI solutions.

HyperVault has previously said it will deploy AMD GPUs within its facilities, and OpenAI has also previously committed to deploying gigawatts worth of AMD hardware.

As well as leasing large amounts of cloud capacity from major hyperscalers, OpenAI is developing data centers in multiple global locations under its Stargate initiative. The company is targeting multiple sites in the US, as well as locations in Norway, the UK, the UAE, Australia, and South Korea, with more planned in Argentina and Canada.

Yotta Data Services to invest $2bn in 20,000 Nvidia Blackwell deployment in Noida, India

Yotta Data Services will invest more than $2 billion to deploy 20,736 Nvidia Blackwell Ultra GPUs at its 60MW D2 data center in Noida, India.

Set to go live in August this year, the ‘supercluster’ will be the first ever deployment of Nvidia B300s in the country.

The cluster will be liquid-cooled, connected via 800Gbps Nvidia Quantum-X800 InfiniBand networking, and supported by 40PB of high-performance parallel file-system storage. More than 10,000 of the B300 GPUs will be committed to the IndiaAI Mission, an initiative launched by the government to support sovereign Indian foundation model development, research institutions, startups, and population-scale public AI platforms.

Within the supercluster, Yotta will also establish the largest Nvidia DGX Cloud cluster in the APAC region as part of a four-year engagement that the company said is valued at more than $1 billion.

The deployment was unveiled during the same week that India is hosting an AI Impact Summit in New Delhi.

In a separate announcement to mark the summit, Netweb Technologies said it would be launching its Tyrone Camarero AI Supercomputing systems built on Nvidia Grace Blackwell architecture. According to the company, the Nvidia GB200 NVL4 platforms will be manufactured in India by Netweb under the government’s 'Make in India' mission, and will feature four Blackwell GPUs and two Grace CPUs to power scientific computing, model training, and inference across the country.

TCS subsidiary HyperVault partners will AMD for 200MW AI buildout across India

HyperVault AI Data Center, the data center subsidiary of Tata Consultancy Services (TCS), is collaborating with AMD to codevelop a rack-scale AI infrastructure design based on the AMD Helios platform.

The news comes a month after the two companies announced a partnership that will see TCS offer AMD Ryzen CPUs, in addition to AMD Epyc CPUs and Instinct GPUs, in an effort to boost AI adoption.

First unveiled in June 2025, AMD’s double-wide Helios AI rack features the company’s Zen 6 Epyc “Venice” CPUs, MI455X GPUs, and Pensando Vulcano NICs, delivering 2.9 exaflops of FP4 performance per rack.

The partnership will also see both companies build out 200MW of AI infrastructure and work with hyperscalers and AI companies to accelerate data center buildouts in India. No timeline for the buildout has been detailed by the partners.

HyperVault was announced during TCS’ Q2 2025 results in October 2025. Though details were sparse, at the time, TCS said HyperVault would oversee the development of 1GW of liquid-cooled data center capacity across India.

Skybox secures approval for data center in Round Rock, Texas

US operator Skybox Datacenters has secured permission to develop a new facility around Austin, Texas.

As reported by CBS and others, the Round Rock city council last week unanimously approved a planned unit development (PUD) from Skybox to rezone just under 30 acres from LI (Light Industrial) to allow for a data center.

The currently vacant land, located east of N. A.W. Grimes Boulevard and south of E. Old Settlers Boulevard, is set to host one data center building totaling up to 250,000 sq ft (23,225 sqm) and 75MW of grid-connected capacity. The site will use a closed-loop cooling system.

Skybox filed plans for the development last year, passing its first reading in December.

A petition from Protect Round Rock against the proposals secured more than 3,300 signatures.

Sabey and Switch are both present around Round Rock, while Amazon has permission for one on land near CR 172 and SH 45. The council also approved in 2024 a 57-acre development near Chisholm Parkway and I-35 that includes a data center, although that project has not begun.

Skybox is a data center builder and operator, providing wholesale colocation, as well as turnkey, enterprise data centers and operations

The company is developing in the Hutto area of Austin alongside Prologis: the PowerCampus Austin project could total 600MW and 4 million sq ft (371,600 sqm). The pair is also developing in the Pflugerville area of the city.

Legrand acquires Brazil's Green4T

Industrial firm Legrand has acquired a Brazilian data center services firm.

As part of its end-of-year earnings announced this week, the company said it had acquired Green4T, a Brazilian specialist in the installation, maintenance, and operation of technical infrastructure for data centers.

Based in São Paulo, Green4T employs nearly 750 people and generates annual sales of around €45 million ($53m). Founded in 2016, Green4T says it has more than 500 data centers and Edge sites under service across Latin America.

This week also saw Legrand acquire Kratos Industries, a US provider of low- and medium-voltage power distribution systems primarily serving data centers.

Legrand is an industrial group headquartered in Limoges, France. The company currently operates in 90 countries, with its offerings including electrical equipment and digital building infrastructure.

The company said it had completed some 30 data center-centric acquisitions to date.

Legrand also recently participated in two-phase direct-to-chip liquid cooling firm Accelsius’ Series B investment round.

The company said data center business represented sales of €2.4 billion ($2.8bn) at year-end 2025, equating to 26 percent of the group’s sales, compared with €700 million ($828m) in 2020.

Meta and AMD sign multi-year agreement for 6GW GPU deployment

Meta has signed a multi-year, multi-generation agreement with AMD for the deployment of up to 6GW-worth of AMD Instinct GPUs.

In a statement, AMD said the partnership will see the two companies align their roadmaps across silicon, systems, and software to deliver AI platforms “purpose-built” for Meta’s workloads.

As part of the deal, AMD has issued Meta a performance-based warrant for up to 160 million shares of AMD common stock, tied to the achievement of specific milestones associated with Instinct GPU shipments.

The first deployment will comprise custom Instinct GPUs based on AMD’s MI450 architecture, with shipments supporting the first gigawatt deployment scheduled to begin in the second half of 2026. The initial deployment will be powered by the custom AMD GPUs and the company’s sixth-generation AMD Epyc CPUs, codenamed ‘Venice’, built on the AMD Helios rack-scale architecture.

Meta will also be a “lead customer” for AMD’s ‘Venice’ CPUs and its ‘Verano’ Epyc CPUs, the latter of which are targeted for 2027.

The agreement closely mimics a deal signed by AMD with OpenAI in October 2025, under the terms of which AMD announced it would also supply the generative AI giant with 6GW-worth of GPUs. That effort is also expected to kick off in the second half of 2026, with OpenAI saying it would build a 1GW data center using AMD MI450 chips.

No information about where Meta or OpenAI will be deploying the initial 1GW has been provided; however, during AMD’s earnings call earlier this month, Su said that with regard to OpenAI, the “ramp starting in the second half of the year going into 2027” is “on track.”

Pacific Northwest National Lab considers building AI data center for DOE research LLMs

The Department of Energy's Pacific Northwest National Laboratory (PNNL) has published a request for information (RFI) for an "on-premises next-gen AI compute data center."

The facility would be funded and operated by the DOE, rather than a separate proposal to allow private data center operators to build AI data centers on Federal land - including at PNNL in Richland, Washington.

In a previously unreported RFI, PNNL contractor Battelle said that it was seeking to understand the data center requirements for "an advanced AI compute infrastructure capable of training and operationalization of multi-data Large Language Models (LLMs) for scientific and national governmental missions."

The system would be around 2MW at launch in 2028, with potential future expansion to ~40MW. It is expected to be capable of supporting training and inference workloads.

RFI questions include understanding data center Tier requirements, UPS configurations, cooling solutions, and the possibility of non-GPU AI chip options.

Vertiv's market cap passes $100bn

Power and cooling firm Vertiv is now worth more than $100 billion.

The company's share price has risen more than 25 percent since it reported robust earnings results on 11 February, and are up nearly 50 percent YTD.

The surge in valuation came ahead of Nvidia's earnings results after the market closed, with those results expected to serve as a bellwether for the AI industry. Vertiv's data center solutions can be used to power and cool Nvidia equipment, and the company is an Nvidia partner.

In its last earnings report, Vertiv reported net sales of $2.88 billion, up 19 percent on the previous year.

With investors increasingly cautious about the long term viability of AI investments, Vertiv reported an order backlog of $15bn, up 109 percent.