Spore.Bio and the EIC Accelerator: Microbe Tests Await Validation

Before a batch of medicine, cosmetics or beverage leaves a factory, someone usually has to wait for bacteria to grow. The standard microbiological test puts a sample on a culture plate and counts the colonies days later: five days in agri-food and fourteen in pharmaceuticals and cosmetics, as Spore.Bio's chief executive Amine Raji put it to TechCrunch in February 2025. Spore.Bio, registered near Paris in December 2022, skips the growth step. It filters the sample onto a membrane, illuminates it at several wavelengths and has a deep learning model read the spectral signature of what it finds.

On 15 June 2026 the European Innovation Council announced 38 companies selected for EIC Accelerator funding, and the published list names Spore Biotechnologies, with the project Unburden, for blended finance. CORDIS now records the grant agreement: signed on 29 September 2026, running from 1 October 2026 to 30 September 2028, with an EU contribution of €2,498,527.50. The project is a month old. What can be checked today is the company's own published data, and those are explicitly preliminary.

Light instead of incubation

The fullest technical description is a whitepaper in European Pharmaceutical Review of 24 March 2026, written by Spore.Bio staff. It names the instrument the LOUIS System and the method TMSI, which the company's website spells out as transformer-based multimodal spectral imaging. A sample is filtered onto a single-use membrane, 8 mm across for simple liquids and 20 mm for complex ones. The membrane is lit at 405 and 365 nanometres and imaged at six wavelengths, picking up the natural fluorescence of molecules such as NADH, flavins and porphyrins. A transformer model built into the hardware turns the images into colony-forming units, the unit a culture plate reports, and was trained to match plate counts.

The company gives seven minutes to a result on the small membrane and 45 minutes on the large one. No dye, reagent or incubation is involved, which is the basis of the "reagent-free" description the company used at its Topeka announcement. The established alternatives are culture plates and a group of rapid microbiological methods that, in the company's own CORDIS summary, are held back by "limited sensitivity, high operational costs, and incomplete regulatory acceptance".

What the whitepaper shows, and what it calls preliminary

The whitepaper follows the approach of PDA Technical Report 33, the industry guidance on validating alternative microbiological methods. In spiked samples of 5 to 300 CFU of Bacillus subtilis, the authors report that LOUIS was non-inferior to plate counting at every level except 25 CFU, where the lower confidence limit was 68.2% against a 70% margin; they consider the gap not significant. Linearity (R²) was 96.50% for B. subtilis and 93.14% for a mix with Candida albicans, and the limit of quantitation 3.5 and 9.0 CFU respectively. In identification, 70 replicates across 23 strains of 11 organisms gave 91.5% correct identifications, 8.5% without an identification and none wrong.

The same document sets out its limits. The data are described as "preliminary results from in-process design experiments", full primary validation is named as a future step, no limit of detection is reported, the identification replicates were run on one day, on one machine, by one operator, and identification works only for strains already in the spectral library. No formal validation under the pharmacopoeia chapters the paper itself cites, USP <1223> and Ph. Eur. 5.1.6, is reported there, and none could be found in a web search on 4 November 2026. The paper is company-authored: its co-author Michael J. Miller, who chairs the revision of PDA Technical Report 33, joined Spore.Bio as vice president of scientific and regulatory affairs in early 2026.

What Unburden is meant to deliver

The CORDIS objective sets the grant's task: move the system "from validated pilots into large-scale industrial deployment", with pharmaceutical bioburden and sterility testing as the main market. Three work streams are named: industrialising the production of consumables, completing adaptation to Good Manufacturing Practice, and regulatory validation under real operating conditions. CORDIS gives no total project cost, and with the project starting on 1 October 2026 there is no periodic report yet. The EIC list awards blended finance, but no EIC Fund investment in Spore.Bio has been announced, in searches of the EIC news pages and the trade press on 4 November 2026.

1. Working instrumentExistsDescribed in the company's March 2026 whitepaper; predates the grant.
2. Industrialised consumable productionToo earlyNo public information on production volumes.
3. GMP adaptationToo earlyNamed in CORDIS as a project task; no result published.
4. Regulatory validation in operating conditionsToo earlyWhitepaper data described by its authors as preliminary; no USP <1223> or Ph. Eur. 5.1.6 validation published.
5. Large-scale industrial deploymentToo earlyNo named pharmaceutical customer found.

Customers: claims without names

Every customer statement found traces back to the company or its press coverage, and none names a customer. In February 2025 TechCrunch reported several commercial contracts covering up to 200 factories, and tech.eu reported prototypes installed in partner factories and a waitlist. In August 2026 the company told Startland News that its systems were deployed across multiple continents, in food and beverage, pharmaceuticals and cosmetics. Raji spent six years in Nestlé factories, according to both TechCrunch and tech.eu, and Silicon Canals described the product in December 2023 as a real-time pathogen detector for factories making food, drink, pharmaceuticals and cosmetics. The EIC project names pharmaceutical bioburden and sterility testing as its primary focus.

One further claim rests on a single source: tech.eu reported that Spore.Bio had acquired a competitor, Greentropism, for its expertise and intellectual property; no second report of that deal was found. TechCrunch and tech.eu both also describe a partnership with the Institut Pasteur giving access to its bacterial sample bank, and Le Monde du Droit says the Series A was carried out in partnership with the institute.

The money, and where the figures disagree

The private funding is well documented, but the figures do not line up. Silicon Canals reported an €8 million pre-seed round led by LocalGlobe on 12 December 2023, when the company had 12 employees. The Series A in February 2025 was led by Singular: TechCrunch and tech.eu report it as $23 million, while Le Monde du Droit, reporting on the law firm that advised the company, gives €23 million. tech.eu put total funding at $31.3 million after that round. In February 2026 the company gave €29.9 million in its announcement of Miller's appointment, and in August 2026 Startland News reported "more than $35 million". Google named Spore.Bio on 26 January 2026 as one of 12 recipients of its $20 million AI for Science fund, without disclosing the company's share.

Headcount is reported unevenly too: 25 employees according to tech.eu and 30 according to TechCrunch in the same week of February 2025, with a plan to reach 50 by the end of that year. The French company register, via the government's business directory, puts the company in the 20 to 49 employee bracket for 2024, lists it as active, and shows no financial figures. No later staff number from a primary source was found.

Where Spore.Bio stands today

The company is active and adding sites. In August 2026 it became the first laboratory tenant of Link Innovation Labs in Topeka, Kansas, reported by WIBW on 5 August and by Startland News on 10 August, with Raji citing "proximity to our customers and deep industrial roots". Neither report gives a number of jobs or an investment figure. The register lists Raji as president and Mistretta and Tazi as managing directors, with Ernst & Young Audit as statutory auditor.

Why it fits European priorities

The Commission's strategy of 2 July 2025 sets the aim of making the EU the most attractive place in the world for life sciences by 2030. The Unburden summary on CORDIS pitches faster product release for manufacturers of pharmaceuticals, food, beverages and cosmetics, and TechCrunch reported interest from developers of cell and gene therapies with shelf lives as short as seven days. CORDIS assigns Unburden 0% on all of its policy trackers, including artificial intelligence, although the method rests on a deep learning model. The company is French, but its newest site is in the United States, and its investors include US funds such as Point72 Ventures.

Spore.Bio was not asked for comment for this article. The reporting is based on publicly available sources, listed below.