Enterprise AI has moved past the question of whether to adopt it. The harder question now is where to run it. As more organizations push generative AI models into production, inference has become the workload that matters most on a daily basis. And for companies in regulated industries, running that inference usually forces an uncomfortable tradeoff: send sensitive data to an outside GPU provider, or slow down the pace of AI adoption to keep that data in place.
Today, Cirrascale and Telehouse France announced a partnership to deploy the Cirrascale Inference Platform directly inside Telehouse data centers. Organizations already hosted in Telehouse facilities can now run AI workloads next to the data those environments already hold, with no need to move that data anywhere.
The problem with distance
Most enterprise inference today happens somewhere other than where the data lives. That distance introduces three recurring headaches. Latency climbs as data travels to external infrastructure and back. Control weakens the moment sensitive information leaves a trusted environment. And compliance gets complicated fast, especially for banks and other regulated businesses operating under European data sovereignty rules.
Cirrascale built its inference approach around three ideas that directly address this: privacy, proximity, and performance. This partnership is what those ideas look like in practice.
Proximity, without the migration
Under the partnership, Telehouse provides the data center and interconnection infrastructure. Cirrascale deploys, manages, and operates the Inference Platform on top of it. For customers, that division of labor is invisible. They gain access to high-performance inference resources without changing their hosting architecture or leaving the colocation environment they already run in.
The platform supports open-source models, proprietary models that customers build themselves, and specialized setups for industries working under strict regulatory constraints. In every case, the data stays put and the customer keeps full control.
As Dave Driggers, CEO of Cirrascale, put it: organizations keep telling us they want to run inference as close to their data as possible, without moving it or giving up control over it. This partnership lets us deliver exactly that, inside the data centers where that data already lives.
Sami Slim, CEO of Telehouse France, framed the bigger shift this way: the future of enterprise AI will be shaped by where data resides. Bringing compute to the environments that host enterprise data lets organizations tap the full value of AI without trading away performance, security, or compliance.
Try it in Paris
To let organizations see the benefits for themselves, a proof-of-concept environment launches in Paris in July 2026. Customers can test the Cirrascale Inference Platform under real conditions, explore new use cases, and evaluate the proximity, performance, and security that come with running inference inside a Telehouse facility.
If you want to talk through what this could mean for your workloads, meet Cirrascale and Telehouse at the RAISE Summit on July 8 and 9, 2026, at booths #3C and #3E. Enterprise teams interested in evaluating the platform can sign up at cirrascale.com/inference-eu.
