Five EIC-backed AI deep tech startups at Intel and Hugging Face Summit in Amsterdam

Brussels, November 22nd 2023
Summary
  • Five startups supported by the European Innovation Council took part in the Intel and Hugging Face Summit in Amsterdam on 22 November 2023.
  • The companies span processor design for exascale, AI accelerators for edge devices, processor IP and tooling, quantum computing for optimisation, and AI for logistics.
  • The appearance underscores the EIC's role in promoting European deep tech but also highlights persistent commercialization and supply chain challenges for hardware and advanced compute.
  • Engagement with Intel and Hugging Face offered visibility and technical exchange but does not guarantee market traction or larger financing.

EIC-backed startups join Intel and Hugging Face Summit in Amsterdam

On 22 November 2023, five European Innovation Council funded start-ups took part in the Intel and Hugging Face Summit hosted in association with Leaseweb in Amsterdam. The summit brought together companies and experts from across the AI hardware and software stack to showcase technologies, exchange technical know how and connect with investors and partners. Organisers billed the event as an opportunity for start-ups to learn from two leading industry players in AI hardware and software acceleration.

Who attended and what they do

CompanyCountryCore focus (as stated)
SiPearlFranceDesigns and markets high performance, energy efficient and secure processors for the European exascale supercomputers project
Axelera AINetherlandsDevelops AI acceleration technology to speed up computer vision and generative AI in edge devices
CodasipCzechiaProvides automated tools and intellectual property for processor design and customisation
Multiverse ComputingSpainUses quantum computing and AI to optimise solutions across verticals and for the green transition and financial and social impact problems
TransmetricsBulgariaUses AI and big data to optimise logistics and transport operations

Why the presence of EIC-backed firms matters

The European Innovation Council supports deep tech companies through grants, business acceleration services and, in some cases, equity co-investment via the EIC Fund. Having EIC beneficiaries present at an industry summit with Intel and Hugging Face is part visibility and part ecosystem building. It signals that these ventures have passed a selection process and are considered strategically interesting by an EU innovation programme. The practical benefits include potential commercial introductions, technical feedback, and investor meetings. However this kind of exposure is an early stage milestone rather than proof of commercial scale or technological sovereignty.

EIC Accelerator and EIC Fund:The EIC Accelerator is an EU instrument that combines grants and blended financing to help deep tech companies scale. The EIC Fund is the equity arm that co-invests in scale ups alongside private investors. Together they are designed to reduce the funding gap for high risk, capital intensive technologies but they do not remove market or supply chain risks faced by hardware projects.

Technical highlights and context

The five companies at the summit represent distinct layers of the compute stack and applications. Their work touches on processor chips for supercomputing, specialised accelerators for edge AI, processor IP and design automation, quantum and AI hybrid optimisation, and operational AI for logistics. Each area has different technical and commercial hurdles. Hardware projects face manufacturing and supply chain constraints. Quantum approaches must demonstrate clear advantage on practical optimisation problems. Software and services must prove they deliver measurable operational savings to customers.

Exascale processors:Exascale denotes supercomputers capable of at least one exaFLOP, that is a billion billion floating point operations per second. Processors aimed at exascale require energy efficiency and reliability at scale. Designing such chips is capital intensive and requires a mature supply chain including advanced foundry partners and packaging technologies.
AI accelerators for edge devices:AI accelerators are specialised processors optimised to run machine learning workloads with higher efficiency than general purpose CPUs. Edge accelerators are designed to operate within power and cost constraints on devices at the network edge such as cameras, industrial sensors and mobile devices. Performance, power efficiency and software ecosystem support determine adoption.
Quantum computing for optimisation:Quantum algorithms promise new ways to tackle optimisation problems that underpin energy systems, finance and logistics. Current quantum approaches are mostly hybrid, combining classical optimisation methods with quantum subroutines. The practical advantage is still problem dependent and often limited by noise and scale of current quantum hardware.

What the participating companies bring and caveats

Each company at the summit brings a clearly stated technical claim. Those claims are plausible but require scrutiny as projects move from prototypes to products.

SiPearl

SiPearl, a French company, develops processors intended for the European exascale supercomputers project. Europe has a political interest in building domestic high performance computing capabilities for scientific and industrial sovereignty. Processor design for exascale is expensive and requires long term partnerships with foundries and system integrators. SiPearl is one of several European efforts to reduce dependency on non EU suppliers but channeling EIC support does not eliminate the capital and ecosystem work needed for mass deployment.

Axelera AI

Axelera AI from the Netherlands focuses on AI acceleration for computer vision and generative AI at the edge. Edge AI is a fast growing segment where efficient inference can unlock lower latency and privacy benefits. Success will depend on the company’s silicon performance, developer tooling and partnerships with device makers. Edge deployments typically require certified components and robust software stacks.

Codasip

Codasip provides processor IP and automation tools for custom processor design. Customisable processors can offer efficiency wins by tailoring instruction sets and pipelines for specific workloads. The market for processor IP is competitive and dominated by incumbents who own broad ecosystems. Codasip’s challenge is to demonstrate differentiated IP and to secure customers across Europe and beyond.

Multiverse Computing

Multiverse Computing positions itself at the intersection of quantum computing and AI to provide optimisation and model compression solutions across finance, energy and the green transition. The promise of quantum accelerated optimisation garners attention and public funding but converting algorithmic advantage into production grade software is non trivial. Practical impact depends on matching problems where hybrid quantum approaches outperform classical methods and on access to scalable quantum hardware or effective simulators.

Transmetrics

Transmetrics uses AI and big data to optimise logistics and transport operations. Logistics has short term potential for AI driven efficiency gains such as better route planning and load matching. The main execution risks are data quality across customers, integration with existing transport management systems and proving consistent returns on investment for fleets and shippers.

Limitations, open questions and risks

Public appearances at industry summits are useful for networking but they are not a substitute for addressing deeper obstacles. For hardware focused firms the time to revenue can be long and depends on access to advanced semiconductor manufacturing, test and packaging. Software and quantum firms must show clear, measurable improvements against incumbent solutions to win customers. Funding from EIC and other EU instruments lowers risk but does not remove market risk. There is also the question of scaling talent and navigating standards and procurement processes across EU member states.

EU strategic autonomy and reality:The EU has policy goals to strengthen strategic autonomy in critical technologies. Investing in European hardware design, software stacks and quantum research supports that aim. However strategic autonomy requires not only design capabilities but foundries, materials, packaging and a competitive industrial base. Those elements remain uneven across Europe.

What to watch next

Trackable milestones to assess the commercial progress of these ventures include announced foundry partnerships, shipping of production grade silicon or accelerator modules, customer case studies with quantified savings, successful follow on financing rounds including EIC Fund co investments, and demonstrable quantum advantage on industry relevant problems. Watch also for partnerships between EU ecosystem actors such as national supercomputing centres, industrial adopters and regional investors.

Events like the Intel and Hugging Face Summit amplify visibility for promising technologies supported by the EIC. They are a useful step in ecosystem building but not a guarantee of success. For policymakers and investors the key question remains whether these initiatives translate into resilient supply chains, sustainable commercial businesses and measurable technology sovereignty over the medium term.

Practical note: the summit took place on 22 November 2023 in Amsterdam and was hosted in association with Leaseweb. The five participating EIC-funded start-ups were SiPearl, Axelera AI, Codasip, Multiverse Computing and Transmetrics.