AI Infrastructure & Data Centers

Europe – October 27, 2026 -- Self-hosted LLM infrastructure tends to become the cheaper option above roughly 50 million tokens processed per day, while commercial APIs usually stay more economical below 5 million, according to AppsTek's analysis of when enterprises need a sovereign LLM. The figures are illustrative and should be checked against current provider pricing and actual workload patterns.

AppsTek's third piece in its sovereign AI series argues that the model layer is now a governance decision, not only a technology choice, as generative AI moves into regulated, revenue-bearing workflows.

Global – October 25, 2026 -- Cloudflare is pitching Workers AI as a serverless inference platform that bills only for usage, at $0.011 per thousand neurons, and runs 50+ models across 200+ cities.

Cloudflare targets idle GPU spend with usage-based billing

Cloudflare argues that inference demand is spiky, unlike training. It cites average GPU utilization of only 20-40%, with one-third of organizations below 15%. Workers AI removes hardware commitments, GPU management and capacity planning, so customers pay only for inference they run.

The pricing page lists 10,000 neurons per day alongside the $0.011 per thousand neurons rate.

Platform exposes Kimi K2.6, GLM 4.7 Flash, GPT-OSS-120B and Llama 4 Scout through one API

The catalog includes:

Reston – October 24, 2026 -- Fast Company has named Seekr's SeekrFlow platform to its 2026 Next Big Things in Tech list in the Established Excellence category, citing its ability to make AI explainable and defensible for enterprises and government agencies.

SeekrFlow targets the gap between AI outputs and regulatory defense

Organizations are moving AI into loan approvals, supply chains, critical infrastructure and national security, according to Seekr. Most still cannot show which data drove an output or defend it to a regulator, auditor or oversight body.

AI-Ready Data Engine and verification tools anchor the platform

The SeekrFlow AI-Ready Data Engine converts raw enterprise data into high-quality datasets. Its verification tools trace data influence, score accuracy and let teams contest and correct errors.

London – October 22, 2026 -- Seekr has opened its first European office in Mayfair, London, to serve UK and European demand for AI that customers can govern, audit and run under their own laws. The office builds on the company's existing UK customer base and a growing pipeline across UK defence, FTSE-listed companies and regulated industries.

Seekr hires Palantir's Naief Al-Shibli to lead UK sales

Al-Shibli joins from Palantir as VP and UK General Manager. He will lead sales and grow the UK business. David Church becomes Managing Director of UK operations. Church has more than 40 years in European investment banking and has advised on data centre transactions across Europe.

Forward-deployed engineers will work inside customers' secure environments

The London team includes forward-deployed engineers who operate inside customers' regulated, secure and sovereign environments. Partners across Europe back them.

NEW YORK – October 19, 2026 -- The four largest hyperscalers plan more than $700 billion of capital spending this year, and Egan-Jones Ratings expects mainstream support from both US parties to keep the AI buildout going despite public opposition to local data centers.

Gallup poll shows 71% of Americans oppose a local AI data center

Egan-Jones cites Gallup data showing 71% of Americans oppose a local AI data center, against 53% for a nuclear plant. Its commentary draws a parallel with drug makers, which have stayed unpopular for years yet kept bipartisan backing while leading all other industries in federal lobbying almost every quarter since 2010.

AI has also gained ground at the ballot box. The pro-AI super PAC network Leading the Future has raised $140 million for the midterms.

United States – October 19, 2026 -- Six of the largest hyperscalers are on track to spend more than $700 billion on computing infrastructure this year, about double the record set in 2025, as gains from shrinking transistors fade and performance is bought by building bigger systems.

Brute-force scaling replaces transistor shrinkage as the main source of progress

The spending goes into new data center sites faster than the electrical grid can feed them. Progress now comes mostly from adding chips, memory, power and cooling rather than from smaller transistors. In AI, training compute has doubled roughly every six months since 2010, far faster than Moore's Law allows, and little of that growth comes from shrinking.

Global – October 19, 2026 -- Tech stocks shed over $1 trillion on Friday as investors grew nervous that AI companies may not turn sustainable profits, NBC reported.

Microsoft also pulled out of a giant AI cloud deal in the UK, according to Gizmodo. The report adds to concerns about the economics of large-scale AI infrastructure.

China weighs curbs on overseas access to its AI models

China is considering restricting overseas access to its AI models, Bloomberg reported. Beijing is also looking at curbing the sharing of chip designs, among other measures. Any such limits would affect foreign developers and buyers that rely on Chinese models.

Global – October 19, 2026 -- Lightelligence has presented dOCS, an optical circuit switching (OCS) interconnect module that it says is built on the world's first silicon photonic OCS chip, aimed at AI clusters, GPU superclusters and high-performance computing (HPC) systems.

dOCS integrates optical switching into a QSFP-DD module

The module embeds switching directly in the interconnect. Signals can be routed through external optical ports or through internal cross-connections.

Lightelligence built the chip with its proprietary integrated photonic integrated circuit (PIC) technology. The built-in switching function links xPUs in a QSFP-DD form factor, which the company describes as deployment-friendly.

Shenzhen – October 19, 2026 -- More than half of cloud data center capital expenditure is expected to go to memory and storage, excluding HBM, by 2027, according to a projection by Morgan Stanley's Daniel Yen presented at the 5th GMIF Innovation Summit. The forum, themed "The Future Built on AI Memory and Storage," was co-hosted by the Shenzhen Memory Industry Association (SMIA) and Peking University's School of Integrated Circuits at the Renaissance Shenzhen Bay Hotel.

Yen said global data center capex will stay strong in the coming years. He named six paths to overcome the "memory wall": process migration, design changes, advanced packaging, memory peripherals, PIM/CIM and new materials.

San Jose – October 18, 2026 -- Linktel Technologies says its 1.6T Linear Pluggable Optics (LPO) 2DR4 module runs at power consumption as low as 10W. The company presented the module to data center architects at the 2026 OCP Global Summit as a way to ease power and cooling loads in AI networks.

Linktel's DSP-free 1.6T LPO design targets latency and component bottlenecks

The linear optics architecture cuts signal processing delays, which Linktel describes as critical for high-performance AI scale-out fabrics. Eliminating the optical DSP also removes component choke points, according to the company, supporting scalability for mass deployments.

Linktel reported multi-switch platform interoperability at the early September 2026 1.6T LPO-MSA Plugfest. It describes that result as a foundation for the 200G-per-wavelength ecosystem.