LumiAIres Is Betting Light-Based AI Chips Can Cut Power by 90%

The deeptech startup, now raising a seed round, has pivoted its initial focus to serve satellite and defence integrators where energy efficiency is non-negotiable.

About LumiAIres

Published

For a hardware startup, the most telling detail is often not the technical spec sheet, but the first customer it couldn't land. LumiAIres, a Glasgow-based deeptech company building neuromorphic photonic processors, started with a pitch to hyperscalers and large tech companies. The feedback, according to the company, was a classic enterprise objection: the product was interesting, but the procurement cycle for a novel chip architecture was too long, and the performance benchmarks too unfamiliar [Techscaler, April 2026]. That rejection became a clarifying moment. The team refocused on a buyer for whom their core value proposition,radically lower power consumption and cooling needs,wasn't just a cost savings, but a fundamental constraint.

The Pivot to Harsh Environments

LumiAIres' technology is built on a premise that computing with light, rather than electrons, can drastically reduce the energy required for AI inference. The company claims its photonic processors can cut AI energy usage by 90% without compromising speed or accuracy [Techscaler, retrieved 2026]. While that's a compelling number for any data center operator watching their power bill, it's existential for operations in space, defence, and remote industrial settings. In a satellite, every watt saved on computing extends mission life or frees up power for other instruments. In a forward-deployed defence system, low thermal output means less detectable heat signature and no need for complex, failure-prone cooling apparatus.

This shift in focus is now the company's stated commercial wedge. LumiAIres is actively seeking conversations with satellite manufacturers, space OEMs, mission architects, and defence integrators [LumiAIres, March 2026]. The bet is that these sectors, which have historically tolerated higher costs for specialized, ruggedized components, will pay a premium for a chip that delivers high-performance AI inference under severe power and thermal constraints.

The Team Translating Research

Commercializing photonic computing requires bridging a deep technical moat. The founding team reflects that dual challenge of advanced research and practical productization. CEO Dr Giulia Marcucci spent nine years in academia and startups developing the underlying light-based AI technology before founding LumiAIres [CodeBase case study, November 2026]. Her co-founder, Jeremy Delhommeau, brought a 15-year background as a manufacturing engineer across MedTech and aerospace, tasked with translating the research into a manufacturable product direction [CodeBase case study, November 2026]. A third co-founder, David Gerard Campbell, adds entrepreneurial and strategic experience [FutureScot, November 2026].

The company has since built out its operational capacity, bringing on a Founders Associate for programme delivery and commercial partnerships, and a fractional Chief Marketing Officer [Scottish Council for Development and Industry, October 2026] [RocketReach, retrieved 2026]. Headcount is small, estimated between one and ten employees, which is typical for a capital-intensive hardware startup at this stage [LinkedIn company profile] [Prospeo.io, retrieved 2026].

A Grant-Fueled Path to Seed

Financing for LumiAIres has followed a staged, grant-heavy path common to European deeptech, providing non-dilutive capital to fund early R&D. The company has tapped into Scotland's support ecosystem, receiving awards from Scottish EDGE, support from Scottish Enterprise, and participation in the Higgs Centre incubator and Space Scotland's Pivot into Space programme [CodeBase case study, November 2026]. This foundation supported the development of its prototypes and the pivotal customer discovery phase.

The company has completed a pre-seed round and is now in the process of raising a seed round. One report from November 2026 cites a seed round of $2.3 million, though lead investors are not publicly named [FutureScot, November 2026]. The following table outlines the known funding trajectory.

Stage Amount Key Notes Source
Early Grants ~$12,700 (est.) Scottish EDGE wildcard award and other support [CodeBase case study, November 2026]
Pre-seed Undisclosed Completed; investors not named [CodeBase case study, November 2026]
Seed $2,300,000 Reported as underway or closed [FutureScot, November 2026]

The Realistic Competitive Set

For a procurement officer in aerospace or defence, LumiAIres is not competing with Nvidia for a socket. The realistic evaluation happens against a different set of alternatives. The competition is bifurcated:

  • Incumbent ruggedized compute. Traditional suppliers of radiation-hardened or mil-spec computing modules that may use older, less efficient semiconductor nodes. The trade-off here is proven reliability versus a step-function improvement in performance-per-watt.
  • Emerging low-power architectures. Other startups working on neuromorphic or analog computing chips designed for edge AI. The differentiation for LumiAIres rests on the photonic approach, which claims inherent advantages in speed and energy efficiency for specific linear algebra operations fundamental to AI.
  • In-house development. Large defence primes and satellite operators with the resources to design custom ASICs for their specific needs. LumiAIres' pitch is that its general-purpose photonic AI platform can be adapted faster and at lower cost than a ground-up proprietary design.

The company's ideal customer profile is a systems integrator or OEM in the space or defence supply chain, one that is architecting a new platform,like a reconnaissance satellite or an autonomous drone,and has the latitude to specify a novel processing unit. The budget owner is likely an engineering lead with a strict power budget, not an IT director comparing cloud instances.

The Roadmap and Risks

The next twelve months are critical for moving from promising prototypes and pilot discussions to designed-in wins. The seed funding will presumably be deployed to finalize chip design, engage with potential manufacturing partners, and secure at least one flagship design partnership. The risks on the path are substantial but defined.

  • Design-win timeline. Sales cycles in target sectors are long and involve rigorous qualification testing. A delay in securing a first major design partner could stretch runway.
  • Manufacturing scale. Photonic chips present unique fabrication challenges. Moving from a lab prototype to reliable, volume production at an acceptable yield is a formidable engineering hurdle that lies ahead.
  • Software ecosystem. AI chips live or die by their software stack. Building robust compiler tools and libraries that allow developers to easily port models onto the photonic architecture is a parallel software investment that must keep pace with hardware development.

The company's answer to these risks appears built into its strategy: by targeting niche, high-value applications first, it can potentially command higher prices to fund the complex journey to maturity, and it can work closely with a small number of lead customers to co-develop the solution. For LumiAIres, the harsh environment of space isn't just a market,it's a proving ground where its value proposition is undeniable.

Sources

  1. [CodeBase, November 2026] Building LumiAIres: How Dr Giulia Marcucci Turned Light-Based AI Into... | https://thisiscodebase.com/case-study/building-lumiaires
  2. [Techscaler, April 2026] LumiAIres: Building a deeptech company beyond the lab | https://www.techscaler.co.uk/resources/building-a-deep-tech-company-beyond-the-lab
  3. [FutureScot, November 2026] Health and net zero startups among 11 Scottish firms named in Techscaler’s ‘ones to watch’ | https://futurescot.com/health-and-net-zero-startups-among-11-scottish-firms-named-in-techscalers-ones-to-watch/
  4. [LumiAIres, March 2026] LumiAIres, March 2026 | https://www.linkedin.com/posts/lumiaires-ltd_spacetech-satellitetechnology-photonics-activity-7434659040430809091-TKfR
  5. [Scottish Council for Development and Industry, October 2026] Scottish Council for Development and Industry, October 2026 | https://www.scc.org.uk/events/e-314
  6. [LinkedIn company profile] LumiAIres Ltd | https://www.linkedin.com/company/lumiaires-ltd
  7. [Prospeo.io, retrieved 2026] LumiAIres company profile | (Source retrieved 2026)
  8. [RocketReach, retrieved 2026] LumiAIres Ltd company profile | (Source retrieved 2026)
  9. [Techscaler, retrieved 2026] LumiAIres company description | (Source retrieved 2026)

Read on Startuply.vc