A British deep-tech start-up promising to liberate AI-powered businesses from spiralling cloud bills has secured £1 million in pre-seed funding, in a deal that points to one of the most pressing margin headaches facing the SaaS sector.

Loc.ai, a London-based outfit building so-called “off-cloud” AI infrastructure, has closed the round under the leadership of Fuel Ventures, the prolific early-stage investor founded by Mark Pearson. The capital will be used to accelerate go-to-market efforts among SaaS and desktop software companies that are presently bleeding margin to the per-call billing model imposed by hyperscale cloud providers.

The pitch is straightforward, if technically ambitious. Rather than routing every user request through a remote data centre, and paying a fee to the likes of Amazon, Microsoft or Google for each one, Loc.ai shifts the artificial intelligence workload directly onto the customer’s own kit, be that a laptop, a workstation or dedicated edge hardware. The result, the company argues, is faster performance, far stronger data privacy and, crucially for chief financial officers, predictable fixed costs in place of variable cloud fees that scale unhelpfully with user growth.

Co-founder Joseph Ward did not mince words. “For years, we’ve handed control of our most critical AI infrastructure to companies we don’t own and can’t influence,” he said. “Inference costs keep climbing. Services get switched off without warning. Loc.ai exists so that developers, governments and businesses never have to accept those terms again.”

That message is landing at a moment when the economics of generative AI are coming under serious scrutiny. With AI no longer a bolt-on feature but increasingly the product itself, embedded in meeting tools, writing assistants, customer-support platforms and code copilots, every keystroke can trigger a billable event. For fast-growing software firms, the result is a cost curve that climbs in lock-step with usage, eroding the margin economics that have long underpinned the SaaS model.

Loc.ai is also tapping into Britain’s intensifying push for sovereign AI. Sensitive material, from boardroom transcripts to customer conversations, is at present routinely shipped through third-party cloud APIs sitting outside national jurisdiction. By keeping inference on-device, Loc.ai claims to remove that exposure entirely, leaving customers with full control over where their AI runs and where their data resides.

The technology is being made viable by the rapid maturation of consumer hardware. Modern laptops can now comfortably run open-source models in the seven to thirteen billion parameter range, sufficient, the company says, to power the bulk of enterprise and SaaS use cases without ever phoning home.

Loc.ai was selected for the inaugural cohort of the Google for Startups Accelerator 2025, a programme that has given the team early sight of the ultra-efficient models being designed by Google for consumer devices. That access has shaped the company’s road map and, the founders argue, positioned it for the architectures that will define the next decade rather than those dominating today’s headlines.

Ward and his co-founder Saif Al-Ibadi are not first-time operators. The pair previously built a deep-tech business applying generative design to defence and aerospace engineering, and counted the Ministry of Defence among their clients, delivering the UK’s first generatively designed rocket engine and reportedly slashing design times by more than ninety per cent. Their pedigree in resource-constrained AI has already been put to commercial use through a multi-year contract with B2Space, which has deployed Loc.ai’s agents at the edge of space and cut bandwidth costs by a similar margin.

Mark Pearson of Fuel Ventures said the firm was backing a problem that has fast become impossible to ignore. “Loc.ai is tackling a critical, margin-eroding challenge facing SaaS as AI usage scales,” he said. “Their deep-tech expertise and track record in deploying AI in constrained environments position them strongly to deliver sovereign AI at scale. We’re excited to support Joe and Saif as they help companies regain control over their technology and costs.”

For the CTOs, engineering chiefs and founders Loc.ai is courting, the proposition is simple: convert an unpredictable variable cost into a fixed one, regain control of sensitive data, and stop subsidising the hyperscalers’ growth with their own margin. With the pre-seed round now closed, the company is betting that an increasing share of the British software industry is ready to listen.

Read more:
British deep-tech start-up loc.ai raises £1m to break SMEs free from the cloud’s ‘inference tax’