The Great Acceleration Meets the Cost of Capital

The market is entering a more demanding phase of the AI Supercycle. Until now, investors could largely treat artificial intelligence, defence technology, sovereign infrastructure and advanced semiconductors as variations of the same expansionary trade. Capital was abundant, valuations could anticipate distant earnings, and acceleration itself was sufficient to command a premium. That assumption is being tested.

China's extraordinary reception for memory-chip champion CXMT, SoftBank's ability to assemble a global banking syndicate behind its OpenAI investment, France's decision to replace Palantir with ChapsVision in a strategically sensitive intelligence contract, and record investment by defence companies in military start-ups all point in the same direction: Intelligence is becoming infrastructure—and infrastructure is becoming sovereign.

At the same time, renewed inflation risks and mounting speculation about a Federal Reserve rate increase are changing the financial equation. The next phase will not be financed at yesterday's cost of capital. Markets must therefore distinguish between acceleration supported by strategic necessity and acceleration sustained only by optimistic valuations. This is no longer simply a contest between technology companies. It is becoming a contest between national systems, capital structures and industrial architectures.

1. 01 CXMT: China Has Turned Semiconductor Sovereignty into a Capital-Market Event

   The debut of ChangXin Memory Technologies is more than a spectacular initial public offering. It is a demonstration of China's ability to connect industrial policy, domestic savings, strategic technology and public equity markets around a national objective. CXMT raised approximately $8.6 billion, potentially rising towards $9.8 billion through the overallotment option, in the largest Chinese semiconductor offering on record. Its shares surged by several hundred per cent on their first day, rapidly making the company one of mainland China's most valuable listed enterprises.

   The apparent excess is easy to criticise. The strategic signal is harder to dismiss. Memory is not a peripheral element of the AI stack. DRAM and high-bandwidth memory are essential to AI servers, data centres, smartphones and accelerated computing systems. CXMT already holds a meaningful share of the global DRAM market, while Nomura reportedly expects that share to rise significantly as China substitutes domestically produced memory for imported supply.

   This is the deeper pattern. The semiconductor cycle is becoming inseparable from the sovereignty cycle. China is demonstrating that industrial policy can create capital-market gravity. Domestic investors are being mobilised not merely to fund a company, but to finance technological independence. Yet the valuation also carries a warning. A strategic asset can still become an overpriced financial asset. CXMT faces equipment restrictions, technological gaps in high-bandwidth memory, geopolitical pressure and formidable competitors including Samsung Electronics, SK Hynix and Micron. The company may become strategically indispensable before it becomes economically dominant. Investors must not confuse the two. signal: A strategic asset can still be an overpriced financial asset — indispensable is not the same as dominant.

2. 02 The Federal Reserve Returns as the Constraint on the AI Supercycle

   The most important market development this week may not come from Silicon Valley or Shanghai. It may come from Washington. Renewed energy-price pressures, resilient economic activity and persistent inflation have increased speculation that the Federal Reserve could raise rates again. Recent market pricing has put the probability of an imminent increase at roughly one-third, while expectations of a later move have strengthened. The US ten-year Treasury yield has moved towards 4.7 per cent, with the thirty-year yield above 5 per cent.

   This matters because the AI boom has been priced under two simultaneous assumptions: earnings growth will accelerate dramatically; and the financial system will continue to provide the capital necessary to build the infrastructure behind that growth. A renewed rate increase challenges the second assumption. The AI investment cycle is extraordinarily capital intensive. It requires semiconductors, data centres, electricity generation, transmission capacity, cooling systems, fibre networks, cloud infrastructure and increasingly large corporate balance sheets. Rising rates increase the discount rate applied to distant AI profits; the cost of financing data centres and energy infrastructure; the debt-service burden of leveraged technology investments; the return threshold required by institutional capital; the probability that weaker AI projects will be postponed or abandoned.

   The next stage of the market will therefore separate strategic capital expenditure from speculative capital expenditure. The strongest companies will continue investing because they cannot afford not to. Hyperscalers, sovereign governments and dominant platforms are building the productive infrastructure of the next economy. The weaker participants will discover that an AI announcement is not a financing strategy. This is where the Great Acceleration becomes the Great Selection. signal: Higher rates do not end the cycle — they separate strategic capital expenditure from speculative capital expenditure.

3. 03 SoftBank and OpenAI: Intelligence Is Now Being Financed Like Infrastructure

   SoftBank's financing for its OpenAI investment illustrates the scale of the transformation. A $40 billion bridge loan, reportedly joined by 21 additional lenders, is not conventional venture financing. It belongs to the financial architecture traditionally associated with large acquisitions, infrastructure projects and systemically important corporate transactions. The financing is reported to be one of the largest bridge facilities ever assembled in the Asia-Pacific region.

   The significance is not simply that banks are willing to lend to SoftBank. It is that a private AI company is now capable of attracting a financing structure of almost sovereign scale. This tells us three things. First, the industrialisation of intelligence requires enormous capital concentration. The most advanced models, AI factories and inference platforms are moving beyond the financial capacity of conventional venture capital. Second, global banks are becoming central participants in the AI Supercycle. They are not only financing data-centre assets; they are underwriting the ownership structures behind foundational intelligence platforms. Third, financial risk is migrating alongside technological opportunity. The larger the AI financing structures become, the more the AI cycle becomes connected to bank balance sheets, credit spreads, collateral values and refinancing conditions.

   The market has celebrated the concentration of technological power. It must now price the concentration of financial exposure. OpenAI is no longer merely a technology investment. It is becoming a major node in the global credit system. That is a mark of maturity—but also a source of systemic sensitivity. signal: OpenAI is becoming a major node in the global credit system — a mark of maturity and a source of systemic sensitivity.

4. 04 ChapsVision: Europe Moves from Digital Regulation to Strategic Substitution

   France's decision to transition its domestic intelligence agency from Palantir towards ChapsVision is one of the most consequential European technology signals of the year. The contract itself may be relatively small compared with hyperscaler capital expenditure. Its strategic meaning is considerably larger. France is effectively declaring that certain categories of data analysis, security software and artificial intelligence cannot remain permanently dependent on foreign platforms. The transition is expected to occur gradually because Palantir remains deeply embedded in sensitive operational systems.

   This marks an important change in Europe's technology debate. For years, European sovereignty was expressed primarily through regulation: data protection, competition policy, digital-market rules and risk controls. ChapsVision represents a different model: sovereignty through substitution. Europe cannot regulate its way to technological autonomy. It must create companies capable of replacing imported technology inside critical systems. That requires strategic public procurement; patient growth capital; access to classified and government use cases; consolidation across fragmented national markets; the willingness to scale European champions before they are globally dominant.

   The ChapsVision decision therefore matters beyond France. It provides a potential template for European industrial AI: use state demand to validate domestic technology, create an initial reference market, and then scale the platform across defence, intelligence, infrastructure and regulated industries. The sovereign AI market may not be won first in consumer applications. It may be won inside governments. signal: Europe is shifting from sovereignty through regulation to sovereignty through substitution.

5. 05 Defence Technology: The Primes Are Becoming Venture Capitalists

   The defence industry is undergoing a structural reorganisation. The world's largest defence companies have reportedly participated in a record $4.1 billion of venture-capital rounds involving military start-ups in 2026. Defence and security start-ups have raised almost $40 billion, while total global defence-related transactions have already exceeded $40 billion this year.

   Traditional defence groups are responding to a new operational reality. Modern conflict is increasingly shaped by autonomous drones; distributed sensors; artificial intelligence; software-defined weapons; electronic warfare; lower-cost, rapidly manufactured systems; continuous battlefield adaptation. The old industrial model—long procurement cycles, highly concentrated platforms and multi-decade development programmes—is being forced to coexist with a faster innovation model. The result is not the disappearance of the defence primes. It is their transformation into orchestrators. They are investing in start-ups, creating venture funds, acquiring specialised businesses and integrating external technologies into established military platforms. Research and development expenditure among major defence manufacturers has risen materially since 2021, while Airbus, Lockheed Martin, BAE Systems, Thales and others are expanding their exposure to dual-use and defence technology.

   This is Velocity applied to industrial organisation. The competitive advantage of the future defence company will not rest exclusively on what it owns internally. It will depend on how quickly it can identify, finance, integrate and deploy innovation originating across an external ecosystem. The defence prime is becoming a capital allocator. The start-up is becoming part of the weapons platform. signal: The defence prime is becoming a capital allocator; the start-up is becoming part of the weapons platform.

6. 06 Europe's Space Ambition: From Strategic Anxiety to Investable Assets

   The Exploration Company is reportedly seeking at least $300 million of new capital at a valuation exceeding $2 billion to develop reusable European space capsules. The company aims to create a European alternative in cargo—and eventually human—transport to low Earth orbit. The Brussels-backed Scaleup Europe Fund is among the potential investors, signalling a growing willingness to direct institutional capital towards strategic European technology.

   This is precisely the kind of company Europe has historically struggled to finance. Europe produces scientific capability, engineering talent and early-stage innovation. It then frequently loses scale-up leadership to deeper US capital markets. The Exploration Company offers a test of whether that pattern is changing. Europe does not need a European copy of every successful American company. It does need independent capabilities in domains where access, security and industrial learning create strategic power. Space is one of those domains.

   The critical question is whether Europe will finance the company only through its development phase—or provide sufficient capital, procurement and political support to carry it through commercial scale. A €2 billion valuation is meaningful. A globally competitive European space architecture would be transformative. signal: The test is whether Europe funds only the development phase — or carries a champion through to commercial scale.

7. 07 Economic Surveys Are Losing the Economy They Claim to Measure

   A further signal deserves attention. Traditional consumer and business surveys are increasingly diverging from actual economic activity. US consumer-confidence readings remain close to levels historically associated with recession, even as GDP growth and household spending have remained comparatively resilient. Business surveys have also repeatedly signalled contractions that did not materialise. The immediate explanation includes falling response rates, political polarisation and differences between large companies and smaller firms.

   But there is a more fundamental interpretation. The economy is becoming more concentrated. Aggregate growth can remain positive even while large sections of the population feel economically insecure. Large companies, high-income households and owners of appreciating financial assets can support spending and investment even as smaller businesses and lower-income consumers experience pressure. The survey data may therefore be less useful as a forecasting instrument—but more valuable as evidence of structural divergence.

   Markets should not dismiss the pessimism. They should understand what it measures. The economy can expand statistically while becoming narrower socially. That gap between aggregate resilience and distributed anxiety is politically unstable. It also reinforces the central thesis of the new capital regime: growth is increasingly being produced by a smaller number of companies, technologies, regions and balance sheets. signal: The economy can expand statistically while becoming narrower socially — that gap is politically unstable.

Implications for institutional investors

The five conditions where the strongest opportunities converge.

• Strategic necessity — the asset is important to national, corporate or technological sovereignty.
• Capital access — the company can finance expansion despite higher interest rates.
• Execution Velocity — management can transform capital expenditure into operating capacity quickly.
• Pricing power — demand is supported by scarcity, infrastructure constraints or mission-critical use.
• Cash-flow visibility — the investment case does not depend indefinitely on falling discount rates.

The most vulnerable assets will be those with the opposite characteristics: high capital requirements, weak strategic relevance, distant profits, repeated refinancing needs and no durable customer commitment. Higher rates will not end the AI Supercycle. They will make it more concentrated.

The first phase of the AI boom was defined by technological possibility. The second is being defined by capital intensity. The third will be defined by sovereignty. China is mobilising its markets to finance semiconductor independence. Europe is beginning to use procurement and strategic capital to build autonomous capabilities. Banks are financing AI platforms at infrastructure scale. Defence companies are reorganising themselves around external innovation. But the cost of money is rising. That changes everything. The winners will not be the companies with the largest promises. They will be the systems capable of directing enormous amounts of capital towards strategic technologies—and converting that capital into operating capability before financing conditions tighten further. Velocity is no longer merely the speed of innovation. It is the speed at which nations, companies and capital markets can turn strategic necessity into industrial power. The Great Acceleration is not ending. It is entering its most consequential—and unforgiving—phase.