GPU-WEAR (coordinated, EUR 1.37M) focused on heterogeneous GPUs for wearables/IoT, and LPGPU2 addressed low-power parallel computing on GPUs.
THINK SILICON
Greek SME designing ultra-low-power GPUs for wearable and IoT devices, expanding into 5G edge computing infrastructure.
Their core work
THINK SILICON is a Greek SME that designs ultra-low-power graphics processing units (GPUs) for wearable devices and IoT applications. They specialize in energy-efficient computing hardware, developing GPU architectures that minimize power consumption while maintaining visual processing capability. More recently, they have expanded into 5G network infrastructure, contributing edge computing and low-cost RAN solutions for affordable network deployment.
What they specialise in
LPGPU2 developed power optimisation frameworks with hardware counters, while GPU-WEAR targeted ultra-low power for wearable devices.
Affordable5G project addressed private 5G networks, O-RAN, neutral hosting, and edge computing for low-cost 5G roll-out.
TETRAMAX focused on technology transfer through competence centers, specifically in customized low-energy computing.
How they've shifted over time
THINK SILICON's early H2020 work (2016-2018) was tightly focused on GPU hardware — low-power parallel computing, power optimization frameworks, and visualization for wearable and mobile devices. By 2017-2021, they began channeling this expertise into technology transfer networks, and by 2020 they pivoted toward 5G infrastructure, applying their low-power computing knowledge to edge computing and affordable network deployment. The trajectory shows a company moving from component-level hardware design toward system-level telecommunications infrastructure.
THINK SILICON is moving from specialized GPU chip design toward broader telecom and edge computing applications, suggesting future collaborations will likely involve 5G/6G infrastructure with embedded low-power processing.
How they like to work
THINK SILICON primarily joins consortia as a specialist participant (3 of 4 projects), but they have demonstrated the ability to coordinate — their largest project (GPU-WEAR, EUR 1.37M) was self-led. With 43 unique partners across 23 countries from just 4 projects, they operate in large, diverse consortia and are well-connected despite being a small company. This suggests they are a valued technical contributor that larger partners actively recruit.
Despite only 4 projects, THINK SILICON has built a remarkably broad network of 43 partners across 23 countries, indicating participation in large European consortia with strong geographic diversity. Their reach spans well beyond Southern Europe.
What sets them apart
THINK SILICON occupies a rare niche as an SME that actually designs GPU silicon for ultra-low-power applications — most companies in this space are either large chipmakers or pure software firms. Their combination of hardware-level GPU expertise with growing 5G/edge computing capability makes them a distinctive partner for projects requiring embedded visual processing in constrained environments. For consortium builders, they bring genuine semiconductor design capability from a smaller, more agile organization.
Highlights from their portfolio
- Affordable5GMarks their strategic pivot into 5G telecommunications infrastructure, applying low-power computing expertise to edge computing and O-RAN.
- GPU-WEARTheir largest project (EUR 1.37M) and only coordination role, focused on their core IP: ultra-low-power heterogeneous GPUs for wearables and IoT.
- TETRAMAXA pan-European technology transfer network that positioned THINK SILICON's low-energy computing as a transferable technology asset beyond their own products.