i3D.net logo in blue
i3D.net logo in blue

i3D.net and Positron AI Partner to Deliver EU Sovereign AI Inference as a Service

i3D.net partners with Positron AI
24 June 2026

Launching first in Europe, the new partnership combines Positron’s purpose-built inference hardware with i3D.net’s global infrastructure to deliver superior price-performance and the pricing predictability today’s AI economics demand.

i3D.net and Positron today announced a partnership to deliver production-ready AI Inference at a fraction of the cost — and with a fraction of the volatility — of Hardware is now deployed in i3D.net data centers in Europe, with broader global expansion to follow. By pairing Positron’s purpose built inference hardware with i3D.net’s global infrastructure footprint, the partnership delivers superior price performance compared to conventional GPU-based systems and gives enterprises the kind of transparent, predictable pricing finance teams can actually budget around.

The economics of AI inference have become the defining constraint on enterprise AI adoption. While AI training workloads have stabilized, inference is growing at a far faster pace — and with that growth comes rising, unpredictable costs. Companies running production AI on GPU-based hyperscaler infrastructure face two compounding problems: per-token pricing that swings with every model change and traffic spike, making internal buy-in and budgeting nearly impossible, and rapidly escalating power costs as inference workloads scale. Organizations increasingly need infrastructure that can scale with their token consumption without sacrificing budget predictability.

That’s where i3D.net and Positron come in. Positron’s hardware is engineered specifically for transformer model inference, delivering better price-performance than conventional GPU-based systems and faster inference where it matters most — response latency, sustained throughput, and consistency under load. It’s also markedly more energy-efficient: Positron’s next-generation hardware is designed to deliver up to 5x more tokens per watt on its core workloads than comparable next-generation GPUs, helping turn energy efficiency into an advantage as workloads scale.

Because models live permanently in memory, performance doesn’t degrade at high volumes, even as traffic scales. Inference requests are served through an OpenAI API-compatible endpoint that drops into existing applications without rewrites. i3D.net brings the global infrastructure, network, and operational expertise to deliver inference as a managed service. Together, the two companies give enterprises a credible path to production-grade AI inference: lower cost, predictable pricing, and the deployment flexibility to run workloads where they need to, including in dedicated EU infrastructure for organizations with data residency requirements.

"The conversation in AI has shifted from how you train a model to how you actually run it at scale. And that's exactly where costs are spiraling and becoming impossible to predict. Our customers need inference infrastructure they can trust in production and budget for with confidence. By bringing Positron's hardware onto our global infrastructure, we can deliver a step-change in price-performance with the predictability the enterprise market has been asking for."

Ivo Teel, CEO, i3D.net

"Inference is the bottleneck the entire industry is trying to solve right now — not just in performance, but in cost and power consumption. Positron's hardware was designed from the ground up to address that: purpose-built for AI inference, with performance that holds at scale. i3D.net brings the infrastructure, operational reliability, and EU footprint that enterprise customers require. Together, we're giving organizations a path to production-grade AI inference that doesn't force them to choose between performance, cost, and data residency."

Mitesh Agrawal, CEO, Positron AI
About i3D.net

i3D.net provides managed private cloud and global network infrastructure that gives organizations the sweet spot between performance, control, and cost. Its fully managed operations cover everything from LiveOps to full data center and hardware lifecycle management, keeping platforms at consistent peak performance globally without the operational load. Underpinning it all is a foundation i3D.net fully controls: a global network engineered for high-throughput performance, paired with owned bare-metal hardware at strategically located PoPs — delivering the predictable speed, low latency, and end-to-end reliability that rented infrastructure can’t match. Customers include Acronis, Saber Interactive, Bluesky, and Mainframe Industries.

About Positron AI

Positron AI builds purpose-built hardware and software to make AI inference dramatically cheaper and more energy-efficient. Positron’s shipping product, Atlas, is designed for rapid, scalable deployment, and the company’s next-generation custom silicon, Asimov, targets tape-out toward the end of 2026 with production in early 2027. Positron’s systems are built to serve long-context and next-generation AI workloads with leading economics. Learn more at positron.ai.

View all News