By Priya Raman

Orca Computing has deployed its PT-3 photonic quantum computer at Digital Realty's LHR19 data center in London Docklands, and the system is live. The machine draws 6kW across two standard racks. A single Nvidia DGX H100 node (a high-end AI training server) consumes roughly 10kW by comparison.

The UK-based firm calls the PT-3 "the first quantum computer built for commercial advantage." Treat that claim with caution. No benchmark data accompanied the launch, and no customer results were published. The PT-3 is modular and upgradable, weighs around 800kg (1,764lbs), and operates at room temperature.

That thermal profile matters for operators. The company said the system installs in standard data center environments without specialized floor, power, or connectivity modifications. Again, no third-party verification exists yet.

How fast was the installation?

James Fletcher, head of solutions architecture at Orca, said at LHR19 that installation took three hours. The system became fully operational within 24 hours.

Latency was the other headline. CEO Richard Murray told the press that speed and latency rank among the PT-3's most important features. He said it is "critical" that the system sits inside data centers, because the AI model it handles "does not operate anywhere near as effectively over a remote link."

That argument cuts both ways. It justifies on-premises deployment. It also means the PT-3's value depends on proximity to classical compute, not on standalone quantum capability.

What is the system actually running?

Two projects are live on the PT-3:

• A hybrid quantum-classical extension of Nvidia's Proteina generative AI model for protein design.
• Optimization problems, in partnership with JIJ Inc.

Orca did not disclose performance figures for either project. The company also did not name further customers.

Murray was candid about the machine's scope. "[Orca doesn't] hide away from the fact our system is a special-purpose quantum computer that leans on things photonics does very well," he said at the launch event. "We're leaning into the things that photonics does best in order to deliver performance, rather than position ourselves as a general-purpose quantum computer."

He added: "Photonics is a different platform. It operates differently, and we've decided that rather than ignore those features, to embrace them in order to deliver valuable applications to our customers."

This framing is honest but limiting. A special-purpose machine narrows the addressable market, and the company offered no timeline for general-purpose capability.

What precedes the PT-3?

Orca Computing was founded in 2019 as a spin-out from Oxford University. The predecessor system, the PT-2, launched in 2024. Before the PT-3 launch, the company had deployed around 11 quantum computers on-premises. Customers include Montana State University and the UK's National Quantum Computing Centre.

Digital Realty hosted an Orca PT-2 at its liquid cooling lab at LHR19 before the PT-3 deployment. The PT-3 runs alongside Nvidia hardware at the facility, integrating with existing high-performance computing (HPC) environments.

What is LHR19?

LHR19, also known as Cloud House West, sits at 1 Muirfield Crescent in London Docklands. The data center launched in 2019 and totals 138,800 sq ft (12,895 sqm) across five stories. CBRE Investment Management owns the facility, having acquired it from Aberdeen Group in 2024.

The deployment signals growing operator comfort with quantum hardware in colocation (a model where tenants rent data center space rather than building their own). Whether the PT-3 delivers measurable commercial advantage remains unproven. The claims are specific. The evidence, so far, is not.