AI Infrastructure & Data Centers

Berlin – October 31, 2026 -- Vertiv and NVIDIA modeled up to 30% more cumulative token output in the first 12 months at equal installed capacity. The gain comes from a combined design-and-deployment approach for AI factories, introduced at NVIDIA GTC Berlin. Polar Data Centers is applying the factory-built element to the first 41 MW phase of its multiphase AI data center at Herøya, Norway.

The pitch targets sequential engineering of power, cooling and facility systems, which Vertiv says creates complexity and potential delays. The new approach starts from pre-engineered reference architectures. These are configured to site requirements, virtually validated, then built as factory-made infrastructure.

Global – October 31, 2026 -- CIQ has released Stratum Kubernetes Engine (SKE), an Apache 2.0-licensed distribution that packages the roughly ten upstream projects a Kubernetes cluster needs into one signed file for installing, joining, upgrading and maintaining clusters on Enterprise Linux.

CIQ built SKE to cut the upkeep cost that AI workloads add, since training and inference spread across many systems and capacity moves with demand. The company is the team behind Rocky Linux, Fuzzball, Warewulf and Apptainer.

SKE pins and tests every component as one set

SKE builds Kubernetes, etcd, containerd, runc and kube-vip from unmodified upstream source. It adds Cilium for networking and CoreDNS for cluster DNS. Users download the file for their architecture and run stratum up.

The same file then handles four post-install jobs:

California – October 31, 2026 -- Switching off Gemma 4's default reasoning with a single flag raised accuracy by up to 48 points, cut cost 6x to 8x and made a correct answer up to 19x faster on short verifiable tasks, according to a 27-model local benchmark run on a Strix Halo machine with a power meter attached. The 12B and 26B variants rose from about 52% to 100% on the easy suite.

Each model answered 190 checkable tasks twice, under two different prompts. Electricity was priced at California summer rates of 34 cents per kilowatt-hour off-peak and 58 cents on-peak.

Gemma 4 reasons on 95 of 95 tasks even when told not to

Asked for a one-line answer with no explanation, Gemma 4 emitted reasoning on every task in both prompt conditions. On a 4,096-token budget, the 12B and 26B spent the entire allowance thinking and returned nothing on roughly half the questions.

ABU DHABI – November 01, 2026 -- Context Labs B.V. launched the AI Distillery, a Context Engine that converts an enterprise's operational and business data into a governed, customer-owned Context Package. The package works with any frontier or open-weight model and runs inside the customer's own environment. The product is available now through the Microsoft Marketplace.

Gartner and McKinsey data frame the AI-readiness gap

Gartner predicts organizations will abandon 60% of AI projects unsupported by AI-ready data through 2026. It also expects over 40% of agentic AI projects to be canceled by the end of 2027. McKinsey reports that eight in ten companies cite data limitations as the barrier to scaling AI agents.

COLUMBUS – October 31, 2026 -- Vertiv (NYSE: VRT) has launched a global ecosystem offering that combines an expanded collaboration with Dow and a new partnership with Solenis to manage liquid cooling as a single, integrated thermal system for AI-scale data centers.

The initiative pairs Vertiv's infrastructure, monitoring and global service capabilities with fluid chemistry, coolant and water treatment expertise from the two partners. Dow supplies advanced coolant technologies, while Solenis contributes specialty chemicals and water treatment solutions.

Vertiv extends Dow Coolant Care Network collaboration and adds Solenis

Vertiv expanded its work with Dow through the Dow Coolant Care Network and started a new partnership with Solenis to support end-to-end fluid management and thermal infrastructure services. Both are integrated with Vertiv's monitoring and lifecycle capabilities.

TAIPEI – October 30, 2026 -- Hon Hai Technology Group (Foxconn) (TWSE: 2317) opened its seventh annual Hon Hai Tech Day with a record 321 exhibits, a quarter of them in a Token Factory zone covering components through full system integration for next-generation AI infrastructure.

Foxconn presents NVIDIA DSX POD as a factory-built reference design for Vera Rubin racks

The NVIDIA DSX POD is a standardized, factory-built reference design. Partners can preintegrate NVIDIA Vera Rubin rack-scale AI infrastructure in modular containers and complete critical validation before shipment. After delivery, the units connect to prepared site infrastructure. Foxconn showed the POD publicly for the first time, and it dominated the venue.

Foxconn also tied its vertical integration to the NVIDIA Vera Rubin platform, spanning rack-level AI supercomputing system integration down to sub-micron-level component manufacturing.

REDWOOD CITY – October 29, 2026 -- Equinix raised its full-year 2026 guidance for the third consecutive quarter after record annualized gross bookings of $533 million, up 35% year over year, and third-quarter revenue of $2.571 billion.

Revenue rises 11% as profit measures grow faster

Revenue increased 11% year over year as reported and 12% on a normalized and constant currency basis. Operating income climbed 27% to $600 million. Net income attributable to common stockholders rose 31% to $489 million, or $4.93 per share, up 29%. The result includes a non-recurring fee earned in connection with the atNorth joint venture investment.

Global – October 29, 2026 -- A malicious version of LiteLLM, version 1.107.0, was published to PyPI on October 29, 2026. It bundles a credential-stealing infostealer with a remote access trojan (RAT), according to OX Security. It is the second hijacking of the AI gateway this year, and PyPI has since quarantined the release.

Malware hides in the wheel package, not the source repository

The attacker injected the code only into the wheel package, so a review of the repository or the clean source would show nothing wrong. Stolen data goes to litellm[.]dev, a domain the attacker bought that can pass for a legitimate LiteLLM address. In March, versions 1.82.7 and 1.82.8 were compromised in the TeamPCP campaign, but this attack used different tooling and infrastructure.

San Francisco – September 10, 2026 -- Innventure, Inc. (NASDAQ: INV) shares plunged 55% on August 14, 2026, after the company pulled its DarkNX data-center project from 2026 forecasts and suspended revenue guidance, triggering a securities class action investigation by Hagens Berman.

Stock lost 74% of value amid unraveling data-center claims

Between May 27, 2026 and August 14, 2026, Innventure shares fell $4.795, or 74%, as scrutiny mounted over its subsidiary Accelsius' purported agreement with DarkNX. The deal, which Innventure had described as "the largest two-phase, direct-to-chip deployment to date," involved deploying Accelsius' NeuCool cooling technology across a 300MW AI data center campus in Ontario, Canada.

New York City – September 10, 2026 -- Innventure, Inc. (NASDAQ:INV) faces a securities fraud class action after its stock collapsed 55% on August 14, 2026, following disclosures that its subsidiary Accelsius had pulled a marquee data center project from its bookings.

Innventure allegedly overstated the viability of Accelsius' DarkNX cooling contract

Law firm Bleichmar Fonti & Auld LLP filed the complaint in the U.S. District Court for the Southern District of New York, captioned Labed v. Innventure, Inc. et al., No. 26-cv-07377. The suit alleges Innventure and senior executives misrepresented an agreement between Accelsius and DarkNX to deploy NeuCool direct-to-chip liquid cooling technology across a proposed 300MW AI data center campus in Ontario, Canada. Innventure had cited the deal to project Accelsius reaching cash flow positivity by year-end 2026 and to support revenue growth forecasts.