TL;DR

The Compression Company, founded by two Imperial College London students, has raised $3.4m (£2.5m) led by Long Journey — an early investor in SpaceX, Uber, and Anduril. The startup uses AI to compress satellite data by more than 95% onboard, addressing the fact that just 2% of satellite-recorded data currently makes it back to Earth.

The Bottleneck Problem

More than 5,000 Earth observation satellites are expected to launch over the next decade, nearly three times the number in the previous ten years. These satellites generate vast amounts of data used in climate monitoring, disaster response, defence, agriculture, and logistics.

But satellites operate with limited bandwidth and only brief windows to transmit data during each pass over a ground station. Research suggests 98% of recorded data is delayed, degraded, or discarded entirely.

“There have been huge investments in capturing more data from space, but far less attention paid to how that data actually gets back to Earth,” said co-founder and CEO Michael Stanway. “Until now, the answer has been to launch more satellites. We are taking a different approach.”

Software Over Hardware

The Compression Company’s AI-driven compression runs directly onboard satellites, reducing file sizes so that operators can transmit more useful information from existing hardware. Rather than launching additional satellites, the approach uses software to extract more value from current constellations.

Stanway and co-founder Joe Griffith, the company’s CTO, met while studying neurotechnology at Imperial College London. The business was initially backed by Entrepreneurs First before closing this new round.

“Space has become a data industry, but the ability to move and work with that data has lagged badly behind its generation,” said Lee Jacobs, managing partner at Long Journey.

Looking Forward

For the UK’s growing space technology sector, the investment signals continued international confidence in British deep-tech startups. The Compression Company’s approach — solving infrastructure constraints with AI rather than hardware — reflects a broader pattern across industries where software is becoming the cheaper path to capacity.