The Next Space Race Will Be Won on the Ground

Even today, with space in the news almost daily, most still picture its future in simple if spectacular terms: rockets blasting off at the end of a countdown, colonies on Mars. They’re the set pieces of the space imagination: loud, bright, easy to grasp. The real action is elsewhere, however, in the machinery on which everything depends. The satellite, the unseen workhorse of the modern world, is evolving faster than ever – faster than perhaps any other area of space.

For most of the history of the industry, satellites were designed to endure. They were capital assets, built to sit in orbit for decades, and updated only when fuel ran low or the next generation of hardware finally justified the expense. But that world has gone; and today, the marriage of market pressures and furious advances in technology have compressed commercial life spans to around eight years or ten at the most. Starlink, racing from Gen1 to Gen4 in short order, has set the pace and the standard. Now, if a system cannot evolve and cannot evolve quickly, it risks obsolescence.

What’s driving this trend isn’t just the satellite hardware itself but what is demanded of them. Modern spacecraft can steer thousands of beams in milliseconds, carving up coverage dynamically to meet user demand. The bottleneck, curiously enough, isn’t the hardware, but the human beings trying to choreograph this complex dance with tools built for a different age. This is the problem of orchestration, which at its heart is a traffic-light problem. It’s about deciding who gets capacity, where they get it, and then they get it.

As satellites get ever-more agile and multi-mission constellations become the norm, orchestration becomes harder by orders of magnitude. Today, much of this is still undertaken manually, which, though certainly requiring a great skill and care, simply isn’t scalable. The market is too unforgiving. If this orchestration problem can be confronted and solved, then satellite system efficiency will leap.

By this point I think we could all be forgiven for finding the two little letters A and I a little uninviting, not least because many of the grander promises of what is undoubtedly a defining technology have not yet come to fruition. But in this instance, AI could make a decisive difference. These systems do not run on magic. They need rapid, constant pattern-finding: they need to be able to anticipate jumps in demand and adjust beam position long before a human operator would.

Indeed, it’s just the job for artificial intelligence, which, as in fields like emergency response, can predict with high accuracy what might happen and make real-time adjustments. Here, it could be used to foresee traffic flows. In a sense, this is a progression from a traffic light system to something more like modern air-traffic control. Planes don’t leave their course. They follow highly predictable flight-paths. Once you know this, you can allocate resources more intelligently. It is similar in the case of broadband demand. AI is well-suited to such structured, semi-predictable environments.

The same revolution is happening on board the satellites themselves. For decades, spacecraft collected raw imagery or signal data and beamed it all back to Earth for processing. This made sense when storage and compute in orbit were scarce. It makes far less sense now. On-board processing, driven by AI accelerators, lets satellites analyze data in situ — filtering, classifying, and extracting meaning before transmitting anything home. Instead of sending gigabytes of raw material, they send the 1 or 2 percent that matters. That shift alone cuts costs, frees bandwidth, and improves the speed and precision of decision-making on Earth. In fields like disaster relief, maritime safety, and defense, minutes matter. AI can give you those minutes.

Innovation is not confined to orbit. The factory floor is also ripe for evolution. In car manufacturing, virtual and augmented reality have transformed training, safety, and throughput. Technicians learn to undertake complex tasks by doing them in a virtual space before they ever touch physical hardware. They build confidence, reduce error, and shorten production cycles.

The world of satellite manufacturing, despite its sophistication, has only begun to take full advantage of these tools. The potential gains to be made here are not abstract. We’re talking about far quicker manufacturing, safer training (and so few injuries and absences), and overall, production lines that can scale to meet the surging demand for new constellations. It means much greater productivity.

In sum, then, the commercial space race will not be won by hardware alone. That time is over. Power will lie with those who can exploit the technology now available to us to orchestrate complex networks as fluid, living systems; who can put AI to work where it can add real value; who can bring operations, training, and manufacturing up to the level the market now expects. Here, on terra firma, is the real commercial battleground.

We are beginning a chapter in the story of space that is defined by speed: speed of deployment, speed of iteration, and speed of learning. The companies that embrace this and find their rhythm quickly will be the ones that, to a great extent, define the next decade of commercial space. Those that cling to the old model may limp along for a while, but soon they’ll fall so far behind that catching up may be impossible. VS

Martin Halliwell is a Partner at NewSpace Capital, one of the only private equity firms dedicated to growth investing in the rapidly expanding space economy. Martin is also the former Chief Technology Officer of SES, where he led global technology and R&D from 2011 to 2019.