The Satellite Industry is in a Period of Momentous Transformation The satellite industry is going through a period of momentous transformation with the emergence of new entrants and new technologies in every segment of the value chain. For decades satellite communications have been dominated by a handful of GEO satellite manufacturers, satellite operators and ground segment manufacturers with almost a cottage-industry-like network of service providers and value-added manufacturers (BUCs, LNBs and antennas). This has been a linear and predictable business model with entirely proprietary technologies. We now see the emergence of new Non-Geostationary Orbit (NGSO), or multi orbit players in LEO, MEO and HEO building completely vertically integrated systems. This shift has significantly driven down capacity pricing: the price of satellite bandwidth for data services has dropped 77% over five years according to analysts Novaspace, formerly known as Euroconsult. Starlink, as the first to market, is making waves by disrupting market sectors historically monopolised by the established GEO players such as maritime, aero and enterprise connectivity. Two years ago, the industry would have dismissed Starlink's impact on maritime or aero connectivity segments. The sentiment was that Starlink has ‘no CIR’ (Committed Information Rate) and therefore would not be considered ‘reliable’ for mobile or critical communications. This notion has since been overturned and the naysayers have paid a price with a significant impact to revenues in maritime—the cruise industry in particular—with Starlink now making inroads into aviation and previously inviolable segments like defence. Starlink has also revolutionised satellite manufacturing, leveraging new technologies such as 3D printing to mass-produce satellites at a phenomenal rate, reducing costs to between $250,000 and $500,000 per satellite. The race is on, with Elon Musk’s Starlink trying to acquire as many subscribers as possible before the challengers like Amazon's Kuiper and Telesat's Lightspeed emerge. Forrester's Digital has predicted that SpaceX’s Starlink broadband-by-satellite system is likely to end 2025 with around 8 million customers (it ended 2024 with approximately 5 million), a remarkable growth rate when you consider that each of the leading GEO satellite operators typically have around 25,000 enterprise VSAT terminals activated. We also see the emergence of Small Sat and MicroGEO manufacturers disrupting traditional commercial models with innovations like satellite-as-a-service. This technology provides additional or targeted capacity for defence and government in hotspot areas. Twenty-five years ago, building and launching a satellite would have cost at least two billion USD. Now we see them being built and launched at a fraction of that cost (circa $60 million), reducing the price per gigabit equal to or below fibre. Starlink has also been fundamental to reducing launch costs. In 1981, launch costs were $147k per kilogram of payload. Starlink’s current generation of rockets have brought this down to $2300 and with the introduction of their new Starship rocket, Elon Musk is talking about a price as low as $100 per kilogram. This scale of reduction in launch costs is driving the democratisation of space by allowing new use cases for space to emerge. The satellite industry is also seeing unprecedented consolidation, coopetition and collaboration, creating a range of new offers to consumers, enterprise and governments. Significant transactions include: In April 2024, SES announced its intention to acquire rival Intelsat. If and when this completes, it will be a significant transaction In May 2023, Viasat completed its acquisition of Inmarsat In October 2023, Eutelsat and OneWeb completed their merger transaction In March 2024, prior to the SES announcement, Intelsat extended its partnership with competitor Eutelsat-OneWeb for LEO services.

In a previous article, Building AI-ready Infrastructure, we looked at the challenges that face the builders of digital infrastructure to create the massive engines that will power the ‘AI Revolution’ – in particular, the mega-data centres that will host the training systems used in Generative AI platforms like ChatGPT.  Most of the attention in the data centre industry is on these monsters, but there is more to it that we need to consider. This article looks at the other uses, applications, and implications of AI, and the infrastructure required to maintain them. The Growth of Industrial AI There are many flavours of AI, and although much of the current focus is on Generative AI, commercial applications use all sorts of other techniques to get the benefits that AI can offer. Indeed, there are some AI experts who think that too much emphasis is being given to the prominent large language models, and that the market will require a more diverse model for deploying infrastructure that will support real-world applications. There are many examples of industrial and manufacturing applications using AI already to optimise, for example, production-line efficiency in factories. These systems take data from sensors and devices (e.g. cameras), and then control the manufacturing processes in real time to improve efficiency, or to reduce the use of raw ingredients – a great example being the use of specialist glues in the automobile industry for sticking windscreens to car bodies – an AI platform has been in use to reduce the amount of glue used without compromising the efficacy of the bond. This may sound, trivial but the quantities used globally mean that even small proportional savings can amount to huge monetary savings. This type of application, used across multiple industries, has enormous potential for saving precious resources (or money), and many industries have been using these techniques for years. However, it is mostly the large manufacturers and processing companies that have been able to exploit this. Deploying this type of system can be expensive and usually entails situating a lot of processing power close to the production line. This excludes smaller enterprises from being able to take advantage as the barrier to entry is too high and involves maintaining IT kit that is expensive and difficult to look after.

Artificial Intelligence (AI) is the hottest topic in technology for many reasons, good and bad, but it’s happening and it’s here to stay, so how do we build the infrastructure necessary to support it? To start with, we should recognise that there are many forms of AI. The one that has created the most buzz is generative AI, as seen in ChatGPT, Meta's LLaMA, Claude, Google’s Gemini, and others. Generative AI relies on LLMs (Large Language Models) which have to be trained using vast amounts of data. These LLMs sit in data centres around the world, interconnected by vast fibre networks. The data centre industry has not stopped talking about AI for at least 18 months, as it gears up for an ‘explosion’ in demand for new capacity. Some of the most respected voices in technology have predicted immense amounts of growth in data centre requirements, with predictions of triple the current capacity within 10 years being at the conservative end. That’s three times the current global data centre market, which has taken 30 years or more to get to where it is today. And, when we say growth, we’re talking about power. AI systems will require three times more electricity than data centres currently consume. Depending on who you ask, that’s about 2-4% of today’s global electricity production. And we’re talking about tripling that, or more. Data Centres So, what is ‘AI-ready infrastructure’ and how are we going to build it? The two key elements are data centres (to house the AI systems) and networks (to connect them with the rest of the world). LLM training typically uses servers with GPUs (the chip of choice for AI) and, for various technical reasons, these work best when in close physical proximity to each other – in other words, GPUs work best in large numbers in large data centres. Not just that, but the new generations of GPUs work best in dense data centres, meaning that each rack or cabinet of AI kit needs a lot of power. Most data centres are designed to accommodate older kit that is not so power hungry. The average consumption globally is about 8kW per rack, although many still operate at about 2kW per rack. The latest nVidia (the leading GPU manufacturer) array needs a colossal 120kW per rack. The infrastructure inside a data centre designed for these beasts is complex: the cooling systems (GPUs run very hot) and electrical distribution systems are much harder to design and set up, and are also expensive. So, data centres for AI training systems are mostly going to be new, as adapting older facilities is a non-starter. So, where do you put them? Finding land next to the vast amounts of electricity required is increasingly difficult in many European countries, especially in the UK. Most of the utility grids in Europe are severely lacking in spare capacity, and building new grid connections and electricity generation is a slow and expensive process. The answer might be to locate these new AI data centres near new renewable energy generation sites, but those are few and far between, so land with access to power now carries a hefty premium. Small nuclear reactors could also be an answer but might take a few years to materialise – we know how to build them (witness the nuclear submarine industry) but getting planning permission to put them on land is another matter. All in all, the data centre industry seems to be at least a few years away from being able to provide the massive upgrade in capacity that is expected. Even solving the land/power problem leaves the issue of actually building a new scale of data centre, 10 or 20 times bigger than what most would consider to be a gigantic site today. It can be done, we can solve the engineering challenges, but these are huge construction projects. Networks What about the networks? Actually, although very little real research has been done on the impact of large-scale AI rollouts on existing networks, we might be in a better position. The fibre networks in the UK and many European countries have benefited from significant investment over the last few years, so coverage is a lot better than it used to be. That does not mean that fast and large fibre routes, which will be a necessity for most AI systems, are all there, but it will be easier to build out new capacity than it will be to find power. Still, what we really need is some serious research into the amount of data that will need to be moved about and how that maps with existing network infrastructure. All in all, we have more questions than answers. Some people in the infrastructure industry are sceptical that things will ever get to the scale that some are predicting, but most of us do expect it to happen – it’s just a matter of time, and the race has already begun. Cambridge Management Consulting Duncan Clubb is a Senior Partner at Cambridge Management Consulting, specialising in data centre and edge compute strategy. Duncan has extensive experience as an IT consultant and practitioner and has worked with many leading organisations in the financial, oil and gas, retail, and healthcare sectors. He is widely regarded as a leading expert and is a regular speaker at industry events. If you or your organisation require support preparing your Digital Infrastructure for the emerging AI-industry, you can read about our array of Data Centre services, and get in touch with Duncan Clubb, through our designated Telecoms, Media, and Technology service page.