AI Infrastructure

Everpure Unveils AI Data Tools Cutting Inference Time 20x

London – – September 30, 2026 -- Everpure (NYSE: P) has launched new platform capabilities that cut large language model inference latency by up to 20x and give enterprises native access to governed data for AI agents, the storage and data management company announced. The updates, rolling out in October, extend the company's Data Primacy architecture first introduced at Pure//Accelerate in June.

PureKVA accelerator slashes AI response times by up to 20x

Everpure's FlashBlade system now pre-stages context directly into GPU memory through a new Key-Value Accelerator, delivering up to 20x faster Time to First Token. The company said the approach supports multi-tenancy without relocating datasets, eliminating GPU idle time and reducing response lag for real-time applications.

SWI Group Reports €4.4B Assets, Pivots to AI Compute Platform

Singapore – September 30, 2026 -- SWI Capital Holding Ltd. (Euronext Amsterdam: SWICH) reported €4.4 billion in total assets and €2.3 billion in Adjusted NAV for the six months ended June 30, 2026, a 53% increase from December 31, 2025, as the group accelerates its pivot from a diversified holding company into a global AI infrastructure and compute platform.

Profit surges on Genesis Digital Assets stake

Profit for the period reached €631.6 million, driven primarily by the value recognized on SWI Group's investment in Genesis Digital Assets, since rebranded SWI Digital. The interim report was reviewed by Deloitte.

Compal Unveils 800VDC AI Factory Infrastructure at OCP Summit

San Jose – September 29, 2026 -- Compal Electronics (TWSE: 2324) will showcase megawatt-scale power and liquid-cooling infrastructure for NVIDIA accelerated computing at OCP Global Summit 2026, positioning rack-level integration as the next bottleneck for AI data centers.

Compal advances 800VDC power architecture to cut conversion losses at scale

The company is developing 800VDC power architecture and megawatt-class power infrastructure aimed at reducing power conversion complexity in next-generation AI data centers. The shift to higher-voltage DC systems targets the rising rack power density driven by accelerated computing deployments.

Plug and Play Selects 140 Startups for Fall 2026 Batches

Sunnyvale, Calif. – September 29, 2026 -- Plug and Play has selected 140 startups from 20 countries for its Silicon Valley Fall 2026 batches, with 55% of participants based outside the United States. The startups will pitch investors and corporate partners at the Silicon Valley Summit, scheduled for Nov. 3-5, 2026.

South Korea supplies nearly a third of the international cohort

South Korea accounts for approximately 30% of companies in the batches, the largest group outside the U.S. Additional startups join from markets across Asia, Europe, the Middle East, North America, and South America, spanning 13 programs including Agtech, Fintech, Insurtech, Health & Longevity, and Travel & Hospitality.

MagPro Debuts 1.85 MW AeroLev Chiller for Water-Scarce AI Data Centers

Singapore – September 29, 2026 -- MagPro, operating as Nanjing CIPU Technology Co., Ltd., has launched the AeroLev 1850, a 1.85 MW-class air-cooled magnetic-bearing chiller designed to cut water use in AI data centers built in arid, high-ambient regions.

Platform eliminates cooling towers to remove water dependency

AeroLev 1850 uses a cooling-tower-free, oil-free magnetic-levitation architecture paired with intelligent controls, removing evaporative cooling water as an operational constraint for facilities in water-stressed markets. The chiller is rated to operate at ambient temperatures up to 45°C.

Castrol Launches CORE, End-to-End Data Centre Cooling Offer

Singapore – September 29, 2026 -- Castrol has launched Castrol CORE, an integrated thermal management offer that shifts the company from selling standalone cooling products to providing end-to-end design, deployment and lifecycle services for data centres.

Castrol moves from product sales to full-service thermal management partnerships

Under Castrol CORE, the company will engage customers earlier in data centre infrastructure development, adding design capabilities to its existing deployment, start-up, operation and maintenance support. Castrol says the change responds to industry research showing growing demand for integrated thermal management solutions that cut the complexity of coordinating multiple suppliers, alongside concerns over supply chain disruption and the high cost of cooling-system failure.

SUNON Unveils Chip-to-Rack Liquid Cooling Line for AI Data Centers

Kaohsiung – September 29, 2026 -- Sunonwealth Electric Machine Industry Co., Ltd., known as SUNON, will present a full chip-to-rack liquid cooling portfolio at the OCP Global Summit 2026, running October 12-15 at the San Jose McEnery Convention Center in California.

SUNON targets rising AI chip power densities with a three-tier cooling architecture

Under the banner "Build to Chill," SUNON's booth A19 display spans chip-level, server-level, and rack-level liquid cooling technologies. The company positions the integrated stack to address a range of power densities, operating environments, and system requirements as AI accelerators push thermal loads beyond the limits of traditional air cooling.

Submer Group Launches Corenix, Names Ex-Johnson Controls Exec CEO

Houston – – September 28, 2026 -- Submer Group has launched Corenix, a modular data center infrastructure company targeting neoclouds, hyperscalers, and AI operators, and named former Johnson Controls executive Martin Renkis as chief executive.

Corenix builds AI data center modules on NVIDIA reference designs

Corenix designs, manufactures, and factory-tests complete data center modules covering IT, power, cooling, and networking before shipping them to project sites for installation, commissioning, and operational readiness. The company positions itself at the deployment layer of AI infrastructure, offering a single accountable team from factory integration through on-site go-live.

Salute Hits 20 GW in AI Liquid-Cooling Deployments in Under a Year

New York – September 28, 2026 -- Salute, a member of the NVIDIA Partner Network, has secured customers to support more than 20 gigawatts of AI infrastructure capacity globally, achieving the milestone in less than 12 months since joining the network.

Salute scales liquid-cooling operations to 20 GW of AI infrastructure in under a year

The growth stems from demand for Salute's Direct-to-Chip (DTC) Liquid Cooling Operations Service, which the company positions as the first comprehensive operational framework for high-density, liquid-cooled AI environments. The service combines field-tested operating procedures, specialized training and commissioning expertise to manage coolant chemistry, leak detection, safety protocols and system monitoring at scale.

SuperX Debuts AI Token Platform, AI Apps at Tech Week Singapore

Singapore – September 28, 2026 -- SuperX AI Technology Limited (NASDAQ: SUPX) unveiled two new products, the AI Token Platform and AI Apps, extending its infrastructure business into model access and application workflows, the company announced at Tech Week Singapore 2026.

SuperX opens access to more than 100 AI models from over 10 providers through one API

The AI Token Platform gives businesses unified access to over 100 AI models from more than 10 providers via a single OpenAI-compatible API and SDK, covering text, image, video and audio models. Access is subject to each model provider's terms. Customers can switch between models without building separate integrations for each provider, addressing a common pain point for AI application developers, internet companies and software firms managing multiple vendor accounts.