Blab Tech
AMD Invests 2 B in UK AI, Launches First Pure Photonic AI Network with London Startup Oriole
AMD has just pledged £2 billion to the United Kingdom, a move that signals the first time a major chipmaker is backing a purely photonic AI network.
The investment is tied to a partnership with Oriole Networks, a London‑based startup that has deployed the world’s first commercial AI network built entirely on photonic technology. By routing data through light instead of electrons, the network aims to cut the energy consumption of GPUs and reduce the frequency of hardware downtime.
Oriole’s system replaces conventional electronic interconnects with light‑based data paths. According to the TechFundingNews article, the photonic network can transmit data at speeds that exceed those of traditional silicon‑based links while consuming less power. The startup claims that the network cuts GPU idle time by a significant margin, although the article does not provide specific performance numbers.
The collaboration follows AMD’s earlier moves into photonic hardware. In 2025 the company acquired Enosemi, a developer of indium‑phosphide photonic integrated circuits, and has filed patents describing hybrid packages that combine a silicon processor with a photonic chip and fiber optics. The new network represents the first time a photonic system has been deployed at scale for AI workloads.
The UK investment comes at a time when demand for high‑bandwidth memory and AI acceleration is outpacing supply. Interactive Investor data shows that Micron’s shares have risen more than 200 % year‑to‑date, reflecting the market’s focus on memory for AI servers. AMD’s own stock price, reported by multiple financial outlets, is trading around $466 per share, indicating investor confidence in the company’s AI strategy. The investment also supports research and development of next‑generation AI accelerators in the UK.
Industry analysts note that photonic computing could address two long‑standing bottlenecks in AI hardware: data movement and power consumption. Optical fibers already provide high‑bandwidth links in data centers, but converting data between electronic and optical formats has historically added latency and energy overhead. By embedding photonic interconnects directly into the AI network, Oriole and AMD aim to eliminate these conversions.
The partnership also aligns with UK government initiatives to strengthen domestic AI capabilities. While the TechFundingNews article does not detail any public funding, the £2 billion commitment signals AMD’s willingness to support the country’s AI ecosystem. The network is expected to be used by data‑center operators and AI research institutions that require high‑throughput, low‑latency compute.
AMD’s broader strategy includes expanding its portfolio of AI‑centric silicon. The company’s recent acquisition of Xilinx in 2022 added field‑programmable gate arrays to its product line, and the Enosemi deal adds photonic components. Together, these moves position AMD to offer end‑to‑end solutions that combine programmable logic, high‑performance CPUs, GPUs, and photonic interconnects.
The commercial deployment of a pure photonic AI network is a milestone for the industry. It demonstrates that optical technology can be integrated into mainstream AI workloads, potentially reducing the carbon footprint of large‑scale inference and training. The long‑term impact will depend on how quickly other vendors adopt similar approaches and how the technology scales to larger data‑center deployments.
As of now, the network is operational in a limited pilot configuration. Oriole and AMD have not announced a public launch date for broader commercial availability. The partnership remains a proof‑of‑concept that could influence future AI hardware design, but additional data on performance, cost, and reliability will be needed before it can be widely adopted.
In summary, AMD’s £2 billion investment in the UK and its partnership with Oriole Networks mark a significant step toward integrating photonic technology into AI infrastructure. The initiative reflects growing industry interest in optical computing as a means to overcome bandwidth and power constraints in AI workloads. The outcome of this collaboration will be closely watched by data‑center operators, AI researchers, and semiconductor investors alike.
The investment is tied to a partnership with Oriole Networks, a London‑based startup that has deployed the world’s first commercial AI network built entirely on photonic technology. By routing data through light instead of electrons, the network aims to cut the energy consumption of GPUs and reduce the frequency of hardware downtime.
Oriole’s system replaces conventional electronic interconnects with light‑based data paths. According to the TechFundingNews article, the photonic network can transmit data at speeds that exceed those of traditional silicon‑based links while consuming less power. The startup claims that the network cuts GPU idle time by a significant margin, although the article does not provide specific performance numbers.
The collaboration follows AMD’s earlier moves into photonic hardware. In 2025 the company acquired Enosemi, a developer of indium‑phosphide photonic integrated circuits, and has filed patents describing hybrid packages that combine a silicon processor with a photonic chip and fiber optics. The new network represents the first time a photonic system has been deployed at scale for AI workloads.
The UK investment comes at a time when demand for high‑bandwidth memory and AI acceleration is outpacing supply. Interactive Investor data shows that Micron’s shares have risen more than 200 % year‑to‑date, reflecting the market’s focus on memory for AI servers. AMD’s own stock price, reported by multiple financial outlets, is trading around $466 per share, indicating investor confidence in the company’s AI strategy. The investment also supports research and development of next‑generation AI accelerators in the UK.
Industry analysts note that photonic computing could address two long‑standing bottlenecks in AI hardware: data movement and power consumption. Optical fibers already provide high‑bandwidth links in data centers, but converting data between electronic and optical formats has historically added latency and energy overhead. By embedding photonic interconnects directly into the AI network, Oriole and AMD aim to eliminate these conversions.
The partnership also aligns with UK government initiatives to strengthen domestic AI capabilities. While the TechFundingNews article does not detail any public funding, the £2 billion commitment signals AMD’s willingness to support the country’s AI ecosystem. The network is expected to be used by data‑center operators and AI research institutions that require high‑throughput, low‑latency compute.
AMD’s broader strategy includes expanding its portfolio of AI‑centric silicon. The company’s recent acquisition of Xilinx in 2022 added field‑programmable gate arrays to its product line, and the Enosemi deal adds photonic components. Together, these moves position AMD to offer end‑to‑end solutions that combine programmable logic, high‑performance CPUs, GPUs, and photonic interconnects.
The commercial deployment of a pure photonic AI network is a milestone for the industry. It demonstrates that optical technology can be integrated into mainstream AI workloads, potentially reducing the carbon footprint of large‑scale inference and training. The long‑term impact will depend on how quickly other vendors adopt similar approaches and how the technology scales to larger data‑center deployments.
As of now, the network is operational in a limited pilot configuration. Oriole and AMD have not announced a public launch date for broader commercial availability. The partnership remains a proof‑of‑concept that could influence future AI hardware design, but additional data on performance, cost, and reliability will be needed before it can be widely adopted.
In summary, AMD’s £2 billion investment in the UK and its partnership with Oriole Networks mark a significant step toward integrating photonic technology into AI infrastructure. The initiative reflects growing industry interest in optical computing as a means to overcome bandwidth and power constraints in AI workloads. The outcome of this collaboration will be closely watched by data‑center operators, AI researchers, and semiconductor investors alike.
