Hello,
As we head into the final months of an already tumultuous year, the intersection between geopolitics and business is only growing more complex. We try to distil this complexity in our daily dispatches, but as much as we do our best (or worst) with the puns, there’s only so much distillation you can do without losing much of the flavour.
Today, we’re launching a new occasional guest column, “Invitation to Treat”, which brings the views of the world’s best minds to add back that flavour through deep perspectives on some of the moment’s biggest themes.
Alongside Iran and energy markets, artificial intelligence has undoubtedly been the biggest of these, intersecting virtually every economic and geopolitical question. And, on the eve of another mega-IPO, there are few more important aspects, both economically and geopolitically, than how the market might finance the enormous buildout the hyperscalers have promised.
There’s also no one better to cover it than Lee Hodgkinson, chief strategy officer of the Saudi Tadawul Group, and a former senior executive at the London Stock Exchange, NYSE Euronext, and SIX.
We hope you enjoy it.
Michael
In 1973, the futurist Arthur C. Clarke wrote: “Any sufficiently advanced technology is indistinguishable from magic.” As we see artificial intelligence, computing, data, digital infrastructure, and space forming a new ecosystem today, his observation from over 50 years ago seems particularly insightful[1].
The magic, however, requires an astonishing amount of capital.
The huge amount of computing power needed by artificial intelligence requires semiconductors, data centres, energy and cooling facilities, as well as connectivity that depends on both terrestrial and ever-more space-based networks and infrastructure. For example, SpaceX’s Starlink is currently changing the scale of low-Earth-orbit satellite connectivity, whilst the possibility of moving computing and data processing into orbit (and even to the moon) is being explored, although significant technical and economic challenges mean this remains an emerging research frontier rather than an immediate next step.
Financing all of this will require capital on a scale reminiscent of the great infrastructure transformations of previous eras, from railways in the 19th century and electrification in the 1920’s to contemporary telecommunications and the internet[2]. Bain & Company estimates that the AI industry will need to generate some $6 trillion of annual revenue by 2031 to support the scale of investment currently envisaged across the new ecosystem. To put that number in perspective, $6 trillion exceeds the entire annual economic output of Germany.
So that’s quite some challenge for the finance industry.
However, the implications go beyond the scale of capital required. As technology acquires the capital requirements of infrastructure, the way in which it is financed is changing, creating new financial risks and dependencies across the ecosystem. As economies and states become increasingly dependent upon that infrastructure, questions of who finances, hosts and controls it are becoming increasingly important. In particular, the ability to determine access to critical capabilities, dictate their terms of use or make switching to alternatives difficult, is becoming a potential source of strategic power.
The Return of Public Markets
For most of the last twenty years, private capital enabled several of the most innovative technology companies in the world to achieve exceptional scale before they went public. It was therefore the case that a great deal of value creation took place before the public-market investors could participate. It seems that may now be changing.
SpaceX’s record-breaking IPO in June 2026 valued the company at approximately $1.77 trillion. Anthropic could provide the next major test, with a valuation of around $2 trillion while carrying more than $500 billion in future commitments, principally tied to the computing infrastructure required to develop and operate its AI models. According to a recent interview with Forbes magazine, OpenAI CEO Sam Altman is planning an IPO in 2027, and others may eventually follow.
However, the return of mega-scale technology companies to public markets creates an interesting paradox. Public market investors may regain some of their access to future value creation that had migrated into private markets, only to find their portfolios becoming still more concentrated in the same technology-infrastructure ecosystem. This raises an important question: how much exposure do investors ultimately want to the growing AI and technology-infrastructure ecosystem?
The technology sector, which includes companies like Apple, Microsoft, and Nvidia, already serves up nearly 40% of the S&P 500. It’s worth noting, however, that this excludes tech-driven giants such as Alphabet and Meta (included in the Communication Services category), along with Amazon and Tesla (contained in the Consumer Discretionary sector). The arrival of another wave of trillion-dollar technology and infrastructure companies could intensify the existing concentration challenge, an issue which is already becoming an important consideration for asset allocators, with BlackRock highlighting AI-related concentration as a growing portfolio-management challenge.
Whether individual IPOs occur according to today’s anticipated timetables is less important than the overall direction of travel. The significance lies not simply in their size and scale, but in how they are increasingly breaking down the boundaries between technology and infrastructure. That’s important because infrastructure has traditionally been valued, financed and risk-managed very differently from technology.
From Technology to Infrastructure
Consider a deceptively simple question: what exactly is SpaceX?
Is it an aerospace company, a telecommunications business, a defence contractor, a digital infrastructure provider, a data business or an AI company? Arguably, it is all of the above.
SpaceX therefore embodies the convergence of technology and infrastructure within the company itself, whereas Anthropic reveals the same convergence through external dependency. It sells intelligence, but faces infrastructure-scale financing requirements without necessarily enjoying infrastructure-like cash flows. Its significant long-term commitments for computing and hosting capacity may provide predictable revenues for infrastructure providers such as Amazon, Google and Microsoft, while exposing Anthropic to material fixed costs without any guarantee of corresponding customer revenues. Broadcom, a leading semiconductor and infrastructure technology player, has agreed to offer up to $42 billion in financing to support Anthropic’s computing needs, which shows how the perimeters of supplier, financier, and commercial partner are starting to blur, giving rise to concerns about the circular nature of the ecosystem.
Data centres reinforce the point. KKR, drawing on McKinsey analysis, points to almost $7 trillion of cumulative global data-centre infrastructure capital expenditure by 2030. Power, land, cooling, grid connections and network capacity are increasingly becoming serious constraints, with the economic effects already becoming visible, as AI-related investment contributed roughly one percentage point to US real GDP growth in 2025.
Alison Porter at Janus Henderson distinguishes between a technology theme, which creates opportunities within an industry, and a technology wave, which requires investment across multiple layers of the economy. AI increasingly looks like the latter, because its development is dependent on an expanding stack of models and software, semiconductors, computing power, storage, networks, data centres, electricity, and connectivity. As a result, investing in one area often increases demand and creates limits in others.
What we continue to call a technological revolution is therefore becoming an infrastructure revolution too.
The New Commanding Heights
There is an interesting historical parallel.
As Arnold Kling and Nick Schulz recount in their 2011 essay, Lenin used the expression “commanding heights” in 1922 to describe the critical sectors upon which the wider Soviet economy depended (principally electricity generation, heavy manufacturing, mining and transportation). Setting Lenin’s politics aside, the underlying observation remains interesting in that some infrastructures matter disproportionately because so much other economic activity depends on them.
Historically, major technological revolutions have been accompanied by the development of new infrastructures upon which subsequent economic activity depends. Railways, electrification, telecommunications and the internet all involved massive capital outlays for new infrastructure. Each of them went through phases of speculation and overinvestment, yet in each case infrastructure was also created which enabled economic activity far beyond what the original investors could have imagined.
Perhaps, then, we should ask what the new commanding heights of the global economy of the twenty-first century are?
Any plausible answer will include compute, artificial intelligence, proprietary data, digital networks and, increasingly, space-based infrastructure. As Lenin would surely appreciate, these are not simply fast-growing industries. They are becoming infrastructures upon which other industries, and increasingly nation-states themselves, depend. Once infrastructure creates dependency, the ability to control critical nodes within interconnected systems can become an important source of strategic power.
However, before considering the implications of that dependency, there is a much more immediate question for capital markets: how will all of this actually be financed?
Capital Markets Are Evolving
The sheer scale of the investment required means this transformation is unlikely to be financed through venture capital and public equity alone. Increasingly, we are seeing the financing requirement extending into debt markets.
Citing Morgan Stanley estimates, the Bank of England has most recently calculated that debt issuance linked to AI had reached $450 billion by early September 2026, which, according to Reuters, is roughly twice the amount seen in 2025. However, debt requires a different discipline. Whereas equity investors can tolerate considerable uncertainty in pursuit of extraordinary growth, credit investors look for predictable cash flows, tangible assets, strong counterparties and contractual protection. Recent developments at SpaceX illustrate the tension. Despite its IPO and subsequent $25 billion bond issuance in June, the company is reportedly seeking a further $40 billion in debt financing to purchase Nvidia chips. The announcement prompted a sharp rise in the cost of insuring its debt against default, with five-year credit default swap spreads increasing from around 110 basis points in June to 194 basis points in October.
The challenge, therefore, is determining how to finance the enormous capital requirements of AI, which parts of the ecosystem can support infrastructure-style financing, which remain exposed to technology-sector risks, and where those risks ultimately reside.
For example, data centres are now driving a lot of real estate fund-raising and can support long-duration financing where cash flows are backed by contracts with investment-grade technology companies. However, as AI-related borrowing grows, credit investors are becoming more selective, favouring companies that have tangible infrastructure and more predictable revenues. According to the Financial Times, this process is already in train, as bonds of several major US and data centre ventures have fallen significantly below their issuance price, as concerns around construction delays, financing costs and the risks associated with increasingly complex commercial arrangements begin to bite. The market evolution is already extending beyond conventional financing. The Chicago Mercantile Exchange is planning to introduce standardised futures based on GPU rental prices, which would enable users and providers of computing capacity to hedge their exposure to future price fluctuations in a way similar to how commodity derivatives allow producers and consumers to manage their risk concerning the prices of energy or raw materials.
All of this shows us that the finance industry is learning how to finance intelligence as infrastructure, as computing power itself is evolving from a technology input into an investable, financeable and increasingly tradeable economic resource.
Power
The story is beginning to move beyond capital markets. As economies become ever more dependent on the technology and infrastructure being financed, questions of capital become questions of power.
In States and Markets, the political economist Susan Strange identified knowledge as one of the four structures of power in the global political economy[3]. Arguably, that insight has never been more relevant. The digital economy is changing the way we create and use knowledge and, at the same time, the systems that help generate, process, and share that knowledge are becoming more important than ever. It is therefore unsurprising that governments are becoming increasingly focused on the geopolitical implications.
Farrell and Newman’s concept of ‘weaponised interdependence’ offers a complementary perspective. They demonstrate how asymmetric networks can create powerful advantages for actors controlling essential nodes or chokepoints, particularly where they can restrict access or make the creation of alternatives difficult to establish. As AI, computing and digital infrastructure become progressively interconnected, similar dependencies may create new opportunities for the exercise of strategic power. The importance of these networks therefore lies as much in the dependencies they create as in the physical assets from which they are constructed.
Recent events in Ukraine illustrate how infrastructure dependencies can become sources of strategic leverage. Restrictions on SpaceX’s Starshield satellite communications service disrupted Ukrainian military operations, with Reuters alleging that the decision originated in the White House. Similar vulnerabilities exist in undersea cables, which carry more than 99% of international data traffic. Their concentration along critical routes exposes global communications and financial markets to disruption, while raising important questions about who controls these increasingly strategic networks. The physical vulnerability of this infrastructure is increasingly apparent, with recent NBC News analysis identifying at least 27 attacks affecting data centres across conflicts worldwide.
“Whosoever commands the Sea commands the Trade,” Sir Walter Raleigh is reputed to have said. Britain subsequently built a large part of its global power on that very principle, since the advantages gained from controlling the infrastructure used for commerce were much greater than the value of the ships themselves[4]. The strategic domains have expanded since then, first to air power and then to space.
Space has long played a strategic role, dating back to the German V-2 rocket launched in 1942, and the 1957 Soviet launch of Sputnik-1, the world’s first artificial satellite. During the Cold War, reconnaissance satellites lowered the chances of nuclear surprise and misunderstanding and the 1991 Gulf War showed that space-based technology could play a key role in military operations. Today, however, falling costs and technological advances are broadening access dramatically, with private companies accounting for the majority of missions and more than 20 nations operating space programmes.
NATO now formally recognises space alongside land, sea, air and cyberspace as an operational domain, reflecting the extent to which modern economies and military capabilities depend upon the infrastructure located there. However, the significance of space increasingly derives from the communications, navigation, data and intelligence infrastructure that rockets and satellites enable.
Space has consequently been described as the “ultimate high ground”[5]. Its strategic significance increasingly lies in the infrastructure embedded within it. For those of us who have spent our careers around exchanges and market infrastructure, this is particularly interesting. We are accustomed to thinking about infrastructure as the systems through which markets operate. Increasingly, the same logic can be applied to this emerging technological infrastructure, where control over critical capabilities can shape the technical standards and commercial ‘rules of the game’ under which everyone else participates. As states themselves become increasingly dependent on privately controlled infrastructure, the boundary between commercial influence and political authority may become harder to define. This is structural power, in precisely the sense that Susan Strange described.
Conclusion
The companies being financed today are constructing parts of the infrastructure through which the future global economy will operate. The challenge for capital markets is to understand how the risks and returns of this emerging infrastructure are distributed across the companies that develop, finance and depend upon it.
As this infrastructure becomes increasingly important to economic competitiveness and national power, questions of financing, ownership and control will intersect with questions of national strategy and security. Of particular importance will be who can control access to critical capabilities, determine the terms of their use and make switching to alternatives difficult.
In this context, the distinction between commercial and national strategic infrastructure may become increasingly difficult to delineate. The coming wave of technology listings therefore matters for reasons that extend far beyond IPO league tables or the profits made by their early investors.
The views expressed in this article are my own and do not necessarily reflect those of any organisation with which I am employed or affiliated.
[1] Clarke, A.C. (1973) Profiles of the Future: An Inquiry into the Limits of the Possible. Revised edition. London: Pan Books.
[2] Perez, C. (2002) Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages. Cheltenham: Edward Elgar Publishing.
[3] Strange, S. (1988) States and Markets. London: Pinter Publishers.
[4] Killingray, D. (2004) ‘Introduction: Imperial Seas: Cultural Exchange and Commerce in the British Empire, 1780–1900’, in D. Killingray, M. Lincoln, and N. Rigby (eds.) Maritime Empires: British Imperial Maritime Trade in the Nineteenth Century. Boydell & Brewer, pp. 1–12.
[5] Christopher Coker, Future War (Cambridge, UK: Polity Press, 2015), 65.