One million pounds, a pre-seed round, and a clear promise: one day, a breath sample will be enough to know what’s happening in the lungs. Respiro Diagnostics, founded by Alison Quinn and Theo Issitt, has closed a round led by Zinc Venture Capital and SFC Capital, with participation from Amadeus Capital Partners, Conception X Angel Syndicate, KQ Labs (Francis Crick Institute), and Innovate UK. The technology combines a proprietary collection device with laboratory methods that isolate DNA, RNA and proteins from exhaled air, offering an alternative to tissue biopsies and blood-based liquid biopsies.

Initial proof-of-concept testing in mesothelioma patients has been completed, and the first clinical study in lung cancer is set to begin at Aalborg University Hospital in Denmark in September 2026. Pulmonary hypertension and asthma are also in the pipeline. The startup has not yet disclosed details of its analytical backend, but the path forward raises a less visible yet decisive question: where will the data run?

The breath paradox: non-invasive diagnosis, high-intensity computing

Analyzing biological material from breath is, by definition, a process that generates complex molecular data. Even without invoking LLMs, identifying predictive patterns in DNA and RNA demands heavy computational models — which today almost always run on cloud infrastructure. But in healthcare, the cloud is an uncomfortable fit: GDPR, fear of data breaches, and the need for immediate reporting push towards on-premise architectures. Breath, collected in a clinic, carries a molecular fingerprint that is already sensitive health data; uploading it to external servers multiplies the attack surface and compliance costs. That’s why Respiro’s funding is not just a clinical bet, but a signpost for med-tech infrastructure: tomorrow’s diagnostics will need local hardware capable of fast inference, whether on edge devices or dedicated hospital servers.

Who wins and who loses in the on-premise diagnostic race

If Respiro’s platform becomes a widespread tool, the beneficiaries will be providers of on-premise accelerated computing solutions — from appliances with embedded FPGAs or GPUs to bare-metal systems for labs. Cloud services positioned as the sole pathway for genomic analysis stand to lose ground, at least in jurisdictions with strict privacy rules. It’s no coincidence that many hospitals are already investing in internal mini-datacenters: the ability to keep clinical data within the walls reduces legal and operational TCO, shifting the debate from “can we do it” to “does it pay off to do it here.”

Respiro will have to untangle a non-trivial technical knot: the collection device is physically separate from the lab analysis. Down the road, miniaturizing the entire pipeline onto a diagnostic cart would mean bringing computation on-device, with power, thermal, and robustness constraints reminiscent of industrial edge computing. That scenario remains speculative for now, but pre-seed rounds also serve this purpose: testing the ground for deployment choices that come after clinical validation. For anyone weighing cloud versus on-premise trade-offs in biomedicine, the evolution of startups like Respiro offers a front-row seat.