The Research and Applied AI Summit (RAAIS) is a community for entrepreneurs and researchers who accelerate the science and applications of AI technology. In the run up to our 10th annual event on June 12th 2026 in London, we’re running a series of speaker profiles to shed more light on what you can expect to learn on the day!
At RAAIS we have a focus on translating cutting edge technology and research into production-grade products for real-world problems.
Philip is co-founder and CEO of Starcloud, a company building data centers in space to address one of AI’s most pressing constraints: energy. As model size and compute demand continue to grow, terrestrial data centers are running into hard limits in grid capacity, cooling, land use, and permitting timelines. Starcloud’s thesis is that space offers a different path: solar power, radiative cooling, and a route to scaling compute beyond what is practical on Earth.
From proof of concept to orbital compute
In November 2025, Starcloud launched Starcloud-1, a 60 kilogram satellite carrying the first NVIDIA H100 GPU ever operated in space, delivering roughly 100x more powerful GPU compute than had previously been deployed in orbit. Within weeks, the company achieved two notable firsts: training a GPT-style language model in orbit using NanoGPT, and running Google’s Gemma model in space on a high-powered GPU.
Processing AI workloads in orbit, close to the satellites generating the data, can cut latency from hours to minutes. Synthetic aperture radar satellites, for example, can produce huge volumes of data that are costly and slow to downlink. Analyzing that data in orbit could materially change both speed and cost.
Why space could matter for AI infrastructure
Starcloud argues that orbital data centers could deliver major reductions in both energy cost and emissions over their lifetime, even after accounting for launch. The appeal is straightforward: space-based systems are not constrained by terrestrial grids, and space offers a naturally favorable environment for radiative cooling without the land and water footprint of conventional data centers.
The company’s long-term ambition is a fully solar-powered orbital data center with 5 gigawatts of capacity, large enough to rival major power plants on Earth without requiring land, transmission infrastructure, or connection to a terrestrial grid.
What comes next
Starcloud-2, currently planned for October 2026, is set to be the company’s first commercial mission. It will carry several NVIDIA H100 GPUs alongside NVIDIA Blackwell hardware, with persistent storage and continuous customer access to orbital compute. The mission will also include a cloud platform from Crusoe, making it possible for customers to deploy and operate AI workloads directly from orbit.
Starcloud is backed by Y Combinator, NVIDIA through its Inception program, and investors including NFX and In-Q-Tel.
Philip’s background
Philip is a second-time founder. He previously co-founded Opontia and earlier worked at McKinsey & Company on satellite projects for national space agencies, giving him firsthand exposure to both the potential and the constraints of space infrastructure.
He holds an MPA in National Security and Technology from Harvard University, an MBA from Wharton, and an MA in Applied Mathematics and Theoretical Physics from Columbia University. He is also a CFA charterholder.