Oliver King-Smith

Verification Futures Conference 2025 was a great chip design event this week at the University of Reading, diversifying into important matters of new AI concepts and IP, with some fascinating new angles on improving AI processing. Well done to Mike Bartley and the rest of the team for such an informative forum. I was really pleased to be asked to speak on the subject of how within an established architecture like RISC-V there is plenty of room for start-up innovation with non-linear thinking and problem-solving. I reviewed the rapid growth in demand for AI at the Edge, both embedded AI where we benefit without even knowing AI is helping, and user-engaged (generative) AI where there is an urgent requirement in many applications to move AI out of data centres with uncertain connectivity and security, and into devices that are co-located with the user - devices that are likely to be battery-powered. An unimaginable leap? Not if we tackle each aspect of power consumption in the AI processing environment. So we can massively reduce the power consumption by a combination of optimised system and chip hardware, more efficient software execution and very refined training models. There's certainly the opportunity for collaboration between innovators, and these goals can be accomplished using today's semiconductor processes. At RED Semiconductor we're developing VISC - a flow control engine for front-end or full integration into a microprocessor architecture. Using rich abstracted instructions we cut large repetitive software routines to just a handful of VISC instructions, so our vector engine and re-ordering hardware systems manage the data flow between memory and the CPU's execution units so that the workload and CPU resources are instantaneously matched for optimum CPU processing efficiency . We're already seeing big jumps in performance, with power-hungry and latency-adding memory accesses being slashed and execution clock cycles being fractionalised. And VISC is moving rapidly along its roadmap of hardware and software performance jumps. What's more VISC will work with any processor - we've started with RISC-V and we designed our own core and are working with major core vendors, too. And it will work just as well when integrated with Arm or any other CPU architecture. To talk to me about any of this, contact info@redsemiconductor.com #semiconductors #AI #RISCV

43

UK AI chip startup Fractile plans a £100m investment to expand its UK operations over the next three years, including a new hardware engineering facility in Bristol. Founded in 2022 and emerging from stealth in mid-2024, Fractile is focused on accelerating AI inference through specialised compute architectures. The expansion will scale teams across London and Bristol, with the new Bristol site hosting both hardware engineering and a test lab for software designed to run next-generation AI models faster than today’s infrastructure. The company previously raised $15m in seed funding from investors including Kindred Capital, the NATO Innovation Fund, and Oxford Science Enterprises. Beyond headcount growth, the move signals a deeper push to anchor advanced semiconductor and AI system design capabilities in the UK. With inference performance becoming a key bottleneck for deploying large models at scale, Fractile’s investment aligns closely with national ambitions around sovereign compute, advanced hardware, and reducing reliance on overseas chip ecosystems. #semiconductors #aitech #ukinnovation #deeptech #hardware

22

2 Comments

Opportunity to Acquire Cutting-Edge Transport IP The Intellectual Property Office UK has now published one of Magway’s core patents: GB2638411A – Controlling a discontinuous linear motor. This patent protects Magway’s innovations in linear motor and electromagnetic propulsion technologies, solutions designed for high-capacity, zero-emission, and scalable goods transport. These innovations have direct applications across: 📦 Automated logistics & fulfilment 🏭 Bulk material handling (mining, ports, heavy industry) 🚇 Subsurface and urban transport infrastructure 🌍 Sustainable mobility & net zero projects With Magway now in liquidation, Alvarez & Marsal are coordinating the sale of this IP portfolio, which represents over 7 years of R&D and more than £12 million of investment. Organisations seeking to strengthen their technology base in sustainable logistics, electrification, or next-generation infrastructure should seriously consider this acquisition. If your organisation is interested, please contact me directly and I will connect you with Alvarez & Marsal, who are running the sale process. 🔗 Patent link: https://lnkd.in/dFxtUtSF

44

3 Comments

https://lnkd.in/ej5aDPGJ Making sure that microgravity research remains a priority post-ISS is critical. But for this industry to excel, it cannot solely exist on Space Stations, it must transfer over to autonomous platforms.

19

UK Scientists Unlock Faster Quantum Measurements with No Accuracy Tradeoff Breakthrough Method Uses Ancillary Qubits to Overcome Speed-Accuracy Bottleneck in Quantum Computing One of the persistent dilemmas in quantum computing has been the tradeoff between speed and accuracy—enhancing one typically sacrifices the other. However, researchers at the University of Bristol have developed a novel approach that circumvents this limitation. By employing auxiliary “helper” qubits, known as ancillary qubits, the team has introduced a space-time tradeoff that significantly speeds up quantum measurements without compromising precision. This discovery represents a pivotal advancement in building efficient and reliable quantum technologies. Key Elements of the Breakthrough • The Speed vs. Accuracy Problem In quantum systems, increasing the speed of measurements often results in greater uncertainty or error, while more accurate measurements slow system performance. This bottleneck has been a major hurdle in the development of practical quantum processors and sensors. • Enter Ancillary Qubits The Bristol team’s approach leverages ancillary qubits—additional quantum bits not part of the main computational sequence—to assist in the measurement process. These helper qubits work alongside the system’s primary qubits to pre-process or encode information in a way that enables faster, cleaner measurement extraction. • How the Tradeoff Works Instead of directly sacrificing time for accuracy or vice versa, the researchers created a spatial workaround: they added more qubits (space) to reduce the time it takes to make accurate measurements. This effectively decouples the traditional constraints of quantum measurement, offering a new paradigm for system design. • Compatibility with Existing Platforms The technique is hardware-agnostic and could be integrated into a wide range of quantum architectures, including superconducting circuits, ion traps, and photonic systems. Its adaptability makes it a strong candidate for near-term quantum acceleration efforts. Why It Matters This innovation addresses a core limitation in the practical deployment of quantum systems. As the field races toward quantum advantage and real-world utility, the ability to measure quantum states both rapidly and accurately is essential. Applications in quantum computing, cryptography, and sensing all rely on high-fidelity information extraction. By removing the need to compromise, the Bristol team’s method enhances system reliability and paves the way for faster, more scalable quantum platforms. It not only accelerates performance—it raises the ceiling on what quantum machines can achieve. Black Rhino Protective Services BlackRhinoGrp.com Now scaling.

14

I’m pleased to share that two companies I founded 9N Advance Materials (Graphene) & 9N Technologies 🤖 have now aligned and are entering into a strategic🤝cooperation agreement. This collaboration brings together advanced graphene materials development and 9N Technologies’ IP-based, CMOS-compatible transistor platform under a single, coordinated strategy. 9N Advance Materials (Graphene) will focus on developing and producing the specific graphene material required for 9N Technologies transistor architecture, making this material a clear priority - while continuing ♻️ broader graphene research as part of its longer-term R&D roadmap. In parallel, 9N Technologies continues the development of its ⚛️ novel transistor design, intended for applications across artificial intelligence, satellites, drones and advanced sensing technologies. While the transistor is still progressing through validation, the alignment ensures that materials and components are developed together from the outset. Under this cooperation, when 9N Technologies enters future partnerships, joint ventures or licensing discussions, the technology can be delivered as an integrated package, combining protected transistor intellectual property + the materials required to build it. This approach simplifies engagement for future partners and supports a more efficient path towards commercialisation for both companies. This alignment also supports a broader long-term vision. As the platforms mature and revenue generation begins, a third pillar, 9N AI Materials, will be introduced. This will focus on materials with embedded intelligence, enabling sensing, adaptability and local decision-making at the material level. Initial development will target robotic skin, with applications extending across multiple industries. After years of parallel development, it is rewarding to see materials, components and future system-level ambitions coming together under one aligned strategy. #Technology #TechInnovation #DeepTech #AdvancedMaterials #Graphene #Semiconductor #AIHardware #FutureTechnology #HardwareInnovation #TechLeadership

34

4 Comments

Wave Photonics has launched a platform designed to simplify the way photonic process design kits (PDKs) are integrated, managed and distributed across multiple electronic design automation (EDA) tools and the UK-based silicon photonics rapid prototyping foundry Cornerstone is its first user. Because designers often use different EDA tools, mapping a foundry’s PDK to each system has traditionally required repetitive setup work, while foundries are also tasked with managing updates, documentation and access for each. With the launch of a PDK Management Platform, Wave Photonics claims users can automate these steps, “allowing teams to focus on higher-value work.” Once a foundry uploads its PDK, the system generates simulation-ready S-parameters – using Wave Photonics’ fabrication models if required – produces up-to-date documentation, packages the PDK for each supported EDA tool and hosts it with version tracking and metadata. Foundries can apply fine-grained access controls and track non-disclosure agreements (NDAs) to maintain confidentiality. The platform currently supports Luceda Photonics’ IPKISS, Cadence’s Virtuoso, Siemens’ L-Edit Photonics and the open-source GDSFactory, with Synopsys Inc’ OptoCompiler and Latitude Design Systems’ PIC Studio planned for the future. “For designers, there is a lot of repetitive work required to access, set up and get S-Parameters for a PDK, and maintaining and updating the PDK for different tools can be a burden for fabs,” said Wave Photonics’ CEO James Lee. “With our EDA partners, we’d already built a solution to this for our PDKs and decided to offer this as a service to foundries so they can improve the designer experience and focus their efforts on high-value work and customer support.” “We’re excited to collaborate with Wave Photonics to bring our PDKs to a wider range of PIC design platforms,” said Dr. Emre Kaplan, PDK Manager at Cornerstone. “Together, we’re building a more connected ecosystem that simplifies access, strengthens PDK management and accelerates the development of next-generation photonic devices.” #Photonics #Microelectronics #MicroelectronicsUK https://lnkd.in/eGeuk7dz

63

1 Comment

I spoke today at the UK Atomic Energy Authority's Fusion for Medical Radionuclides event on "Fusion Technologies for Industrial and Medical Applications." At SHINE Technologies, we already use fusion neutrons to test advanced materials. And we'll use them to make the medical isotopes to help treat cancer. These are practical steps that also build the foundation for future fusion energy systems. Our partnership with UKAEA through the LIBRTI program is another example of how fusion technologies can advance both near-term applications and long-term goals.

122

1 Comment