Navigating the AI of the Storm: Opportunities, Risks and Structural Limits

Artificial intelligence has rapidly become one of the most powerful narratives in global markets. Since the launch of ChatGPT, roughly USD 10 trillion in market capitalization has been added to equity markets. Investment in AI infrastructure is accelerating at an extraordinary pace, with capital expenditure expected to reach around USD 600 billion in 2026 among the Magnificent Seven alone. At the same time, AI already accounts for between 1% and 2% of global electricity consumption, a figure that could double by 2030.

Yet behind the technological momentum lies a complex and global ecosystem. AI depends on an intricate supply chain that spans semiconductors, data centers, energy infrastructure, and raw materials, all embedded within geopolitical dynamics. The scale of investment raises fundamental questions: are the macroeconomic benefits sufficient to justify the spending? Can infrastructure keep pace with development? And how resilient is the current concentration of leadership within a small group of technology firms?

In a recent webinar, Emilie Tetard -  Quantitative & Cross Asset Strategist, Christopher Hodge – Chief US Economist, Hadrien Camatte – Economist for France, Belgium and the Euro Area, Thibaut Cuilliere – Head of Sector Research, Eric Benoist – Tech and Data Expert, Trinh Nguyen – Senior Economist for Emerging Asia, Gary Ng – Senior Economist for Sectoral Research – Asia, and Rita Boutros – Tech and Data Analyst, explored the macroeconomic implications and potential market risks.

The current economic cycle is increasingly influenced by technology investment, with AI playing a central role in driving corporate capital expenditure. In recent quarters, spending on AI and related technologies has represented the majority of growth in business investment, particularly in the United States.

This surge in investment has distinct macroeconomic implications depending on the time horizon. In the short term, the construction of data centers and expansion of computing infrastructure is supporting economic growth and job creation. Large-scale infrastructure projects require labour, materials, and capital, contributing to increased economic activity.

QUOTE: “AI investment is driving the current economic cycle, boosting growth in the short term while potentially transforming productivity and labour markets over the longer run.”

Over the longer term, however, the impact could shift significantly. AI-driven productivity gains may ultimately place downward pressure on prices while reshaping labour markets as automation expands across sectors.

Adoption is progressing steadily across firms of all sizes, though early signs suggest that implementation may take time to scale across the broader economy. While AI is expected to become a pervasive technology across industries, the full macroeconomic effects will likely unfold gradually over the coming years.

Over the longer term, however, the impact could shift significantly. AI-driven productivity gains may ultimately place downward pressure on prices while reshaping labour markets as automation expands across sectors.

Adoption is progressing steadily across firms of all sizes, though early signs suggest that implementation may take time to scale across the broader economy. While AI is expected to become a pervasive technology across industries, the full macroeconomic effects will likely unfold gradually over the coming years.

The AI ecosystem is increasingly characterized by deep interdependence between major technology companies, cloud providers, developers, and investors. One company that illustrates this dynamic particularly well is OpenAI, which has become a central player within the global AI network.

OpenAI has raised more than USD 170 billion in private equity funding, with its valuation approaching levels that would make it one of the most valuable companies in the world if publicly listed. Strategic partnerships with major technology firms have further embedded the company within the broader AI ecosystem.

However, the financial structure of the company remains challenging. Despite strong revenue growth, Open AI continues to face extremely high operating costs driven primarily by computing infrastructure. For example, current revenue levels remain offset by similarly large losses, reflecting the enormous cost of training and running advanced models.

Even ambitious growth projections suggest that computing costs could continue to outpace revenue for several years. While subscription models, developer access and emerging AI applications may generate significant income streams, the long-term sustainability of these models is still evolving.

This creates a new form of systemic risk: the ecosystem has become sufficiently interconnected that major disruptions affecting key players could reverberate across the entire AI landscape.

The scale of investment required to build AI infrastructure is unprecedented. Capital expenditure among the 5 largest US hyperscalers (Microsoft, Google, Meta, Amazon, Oracle) could reach around USD 650 billion by 2026, roughly triple the levels observed only 2 years ago.

Despite strong operating cash flow, some companies may temporarily become free-cash-flow (after dividends and share buy-backs) negative in 2026, requiring additional financing through debt markets.

However, credit markets have so far shown strong appetite for those issuance. Recent bond placements by large technology firms have attracted substantial investor demand, highlighting continued confidence in the long-term prospects of the sector.

However, the growing supply of technology bonds is beginning to influence market dynamics. Spreads have widened in secondary markets, reflecting the increasing volume of issuance and investor attention to differences in leverage, liquidity, and exposure to AI investment cycles.

In the coming months, we expect an increasing risk premium for Technology bonds, particularly on longer maturities. Differentiation between companies will also become more pronounced as investors assess which firms are best positioned to manage the financial demands of the AI expansion.

The long-term economic justification for massive AI investment rests on the productivity gains the technology may generate.

Early micro studies show very promising results. For instance, AI tools can significantly improve efficiency in tasks such as customer service, research, analysis, and coding. Workers equipped with AI assistance are often able to complete tasks faster and produce higher-quality results.

However, translating these gains into measurable macroeconomic productivity growth remains uncertain. Historically, technological revolutions often take years before their full impact becomes visible at the aggregate level.

Economic modelling suggests that AI could increase total factor productivity by around 0.1% to 1.2%, a very wide range, but outcomes depend heavily on adoption speed and organizational change within companies. Businesses must not only deploy the technology but also redesign workflows and operating structures to fully capture its benefits.

In Europe, adoption highlights both the promise and the challenge of this transition. Public use of AI tools is rising rapidly – in France, around 44% of the population already uses AI technologies. Yet corporate adoption remains uneven, with large companies integrating AI faster than smaller firms.

Whether AI ultimately leads to job creation or displacement remains an open question, with studies suggesting both outcomes depending on sector and implementation.

Generative AI has the potential for significant productivity gains, but it's not automatic. It hinges on profitable adoption and business adaptation

Asia plays a central role in the global AI ecosystem, particularly through its dominance in semiconductor manufacturing and technology hardware.

Taiwan, South Korea, and Japan are critical suppliers of advanced semiconductors, memory chips and manufacturing equipment required for AI infrastructure. Taiwan alone accounts for more than 75% of global semiconductor foundry capacity, making it indispensable to the global supply chain.

The economic benefits of the AI boom are therefore unevenly distributed across the region. Economies with strong technology export sectors are already seeing significant growth linked to rising global demand for AI hardware.

Taiwan provides a clear example, with GDP growth exceeding 8% in 2025, more than half of which was driven by net exports related to technology demand.

Other parts of Asia are benefiting through investment flows. Countries such as Malaysia are attracting foreign direct investment in semiconductor packaging and data center infrastructure, although these projects tend to be highly capital intensive and generate fewer jobs.

Over time, productivity gains may provide a broader growth channel across the region, but the distribution of benefits is likely to remain uneven.

Global competition is intensifying as countries seek to position themselves within the AI ecosystem. Asia remains essential to the supply chain, but the strategic competition between the United States and China is becoming increasingly important.

China is pursuing a distinct strategy focused on cost efficiency and large-scale deployment of AI applications. While export controls limit access to some advanced computing technologies, Chinese companies are exploring alternative approaches, including using larger quantities of less powerful chips and optimizing algorithms to maintain competitive performance.

China also benefits from significant advantages in data availability and large-scale industrial adoption of AI technologies. These factors allow the country to develop strong capabilities in AI applications, even if it faces constraints in cutting-edge computing power.

Although the United States currently leads in AI platform development, China remains a potential long-term challenger due to the breadth of its technology ecosystem and the scale of its domestic market.

Energy supply is emerging as one of the most important structural challenges for the expansion of AI infrastructure.

Data centers are extremely energy intensive, with electricity representing 20% to 30% of their operating costs. As AI systems grow more powerful, electricity demand from data centers is expected to increase significantly.

Estimates suggest AI-related electricity consumption could reach around 900 terawatt hours annually by 2030, roughly equivalent to Japan’s current electricity consumption.

At the global level, this increase appears manageable, representing around 3% of total electricity demand. However, local constraints are far more significant. Data centers require reliable, continuous electricity supply and must be connected to power grids, which can take years in some regions.

As a result, access to electricity is becoming a key factor in determining where new AI infrastructure can be built. Technology companies are exploring multiple strategies to secure power supply, including direct power purchase agreements, partnerships with energy providers, repurposing coal-fired plants and investing in emerging nuclear technologies.

Regulation is becoming an increasingly important factor shaping the future of AI development.

Different regions are adopting distinct regulatory approaches. The United States has generally pursued a pro-innovation strategy aimed at preserving technological leadership and encouraging rapid development.

Europe has adopted a more precautionary framework focused on managing risks associated with AI systems, particularly in high-risk applications. While this approach may enhance safeguards, it also raises compliance costs and may slow adoption, particularly for smaller companies.

China represents a hybrid model combining strict regulatory oversight with strong government support through subsidies and infrastructure investment. This combination has enabled Chinese firms to scale rapidly despite regulatory constraints.

Ultimately, regulation does not necessarily hinder innovation. The key factor is whether regulatory frameworks are accompanied by sufficient financial support, infrastructure, and policy alignment to sustain technological development.

Artificial intelligence is reshaping the global economy at an extraordinary pace. The technology promises major productivity gains and is already driving unprecedented levels of investment in computing infrastructure, energy systems, and semiconductor production.

Yet the AI ecosystem is also revealing new structural challenges. Energy supply, semiconductor bottlenecks, financial market dynamics, and geopolitical competition will all influence how the industry evolves in the coming years.

Rather than a simple technological revolution, AI is emerging as a complex global system that connects digital innovation with physical infrastructure and economic policy.

The opportunities remain immense, but the trajectory of AI will depend not only on breakthroughs in algorithms and computing power, but also on how successfully economies adapt to the broader structural demands of this new technological era.