Terakraft and Neurophos collaborate on sustainable AI
The companies plan to run a pilot in 2027, offering early access to Neurophos’ ultra-efficient optical chips for AI inference hosted in Terakraft’s fully renewables-powered datacentre
AI datacentre operator Terakraft and AI chip company Neurophos have announced that they are collaborating to provide sustainable, high performance, and energy efficient AI infrastructure.
As the AI-driven surge in demand for data continues, a major challenge is the enormous amounts of energy required to power and cool computing infrastructure. Terakraft and Neurophos seek to demonstrate that a combination of green datacentres and breakthrough hardware can meet these needs in a sustainable way.
Terakraft, which is based in Norway, says its infrastructure is fully powered by renewable hydropower and cooled with natural lake water, achieving a power usage effectiveness (PUE) below 1.1, which it says is among the best metrics in the industry. The company is repurposing an existing hydropower plant built with reinforced concrete, with the goal of avoiding additional embodied carbon and setting a new benchmark for low-emission operations.
Meanwhile, Neurophos has developed AI hardware with proprietary optical processing units which it says are 100x more energy efficient than leading GPUs for the same workload. The company says it has successfully fabricated, demonstrated, and characterised its miniaturisation of optical modulators by a factor of 10,000x, enabling future optical processing units to deliver the compute power of 100 GPUs while consuming the equivalent of 1 percent of the energy, validated by end-to-end simulation results. According to Neurophos, its compute-in-memory architecture, which is inspired by the human brain’s efficiency, allows for unprecedented processing speeds and density, making ultra-efficient, large-scale AI inference practical and scalable.
The companies plan to host a pilot as part of a commercial early access programme in 2027 for Neurophos’ accelerated AI inference platform – a project that aims to provide a real-world proving ground for sustainable, ultra-efficient compute.
“By hosting Neurophos’ ultra-efficient optical chips in our green datacentre for select enterprise clients, we not only reduce our carbon footprint but also raise the bar for energy-efficient AI infrastructure,” said Giorgio Sbriglia, chairman of the board of Terakraft. “Our mission has always been to power the future responsibly, and this collaboration brings that vision to life.”
Patrick Bowen, CEO and founder of Neurophos, added: “Terakraft’s commitment to renewable energy and innovative technologies aligns perfectly with our mission to democratise high-performance AI. By deploying our 100x more efficient inference chips in Terakraft’s green datacentre we’re proving that AI’s exponential growth can be achieved sustainably, together.”
