Artificial intelligence (AI) remains a key theme in the market, but currently, the debate surrounding the AI cycle has become more demanding. After an initial exuberant phase, fueled by ChatGPT and the resurgence of leaders in semiconductors and platforms, more difficult questions are emerging.
Do profits still justify valuations? Is the deployment of AI based on genuine demand or exaggerated enthusiasm? Can the spending of hyperscalers be sustained? Given the significant divergence between semiconductors and software, where should investors position themselves?
These questions formed the central theme of our webinar , “The AI Cycle: Earnings Reality, Bubble Risks, and Preferred Positioning.” Moderated by John Woods (CIO and Head of Investment Solutions for Lombard Odier Asia), the discussion brought together Corinne de Boursetty (Investment Strategist – Sustainability and Research) and Jian Bo Gan (Equity Research Team Head, Asia).
In his introduction, John Woods announced that the session aimed to explore “the perspectives and opportunities shaping today’s AI-driven markets.” The goal was not to amplify the background noise surrounding AI, but to examine this theme in light of the variables most relevant to investors: valuations, adoption, capital expenditures, funding quality, regional differentiation, and portfolio positioning. The speakers collectively explained why AI is “not (yet) a bubble,” why Chinese technology deserves special attention compared to that of the United States, and how to invest across the entire AI value chain.
The AI ecosystem is not limited to this single technology: it encompasses information technologies, communication services, and certain segments of non-essential consumption, together representing approximately 35% to 40% of the global equity market.
A broader view of the markets
Corinne de Boursetty began by broadening the scope of the study. The AI ecosystem, she asserted, is not limited to this single technology: it encompasses information technology, communication services, and certain segments of consumer non-essentials, together representing approximately 35% to 40% of the global equity market. AI is therefore already influencing several of the market’s main sources of return, as demonstrated by the MSCI All Country World (ACWI) index. Since January 2025, communication services have grown by 30% and information technology by 26%, compared to 25% for the MSCI ACWI index as a whole. Consumer non-essentials, meanwhile, remained subdued at 4%, held back by concerns related to tariffs and a potential slowdown in growth.
Even though this divergence has fueled bubble fears, for Corinne de Boursetty, strong stock performance alone is not synonymous with excessive speculation. To incorporate earnings, final demand, and balance sheet quality, the risk of a bubble should instead be assessed against five key pillars: valuation, AI adoption rate, capital expenditures of hyperscalers, financing quality, and energy availability.
“A level far below the peak of the dot-com bubble”
Regarding the first pillar, valuation, the tone was measured: the MSCI US IT index is trading at 23 times projected earnings over twelve months, a level significantly lower than those reached during the dot-com bubble. Meanwhile, global technology company earnings are expected to increase by 39% in 2026, compared to 15% for the overall market. Furthermore, the sector’s price-to-earnings ratio stands at 1.1, compared to 1.4 for the equity market: although valuations are high, they remain supported by superior growth.
Concentration remains high, and the market is still heavily dependent on a small number of leaders, such as Amazon, Microsoft, and Alphabet. However, the data suggests that the current AI cycle rests on much more solid foundations than a mere speculative boom.
Thus, unlike what happened during the dot-com bubble, earnings per share have, so far, generally kept pace with rising stock prices. As Corinne de Boursetty pointed out, “there is a certain dichotomy between the recent pullback and the maintenance of very solid fundamentals,” adding that “the current level is well below the peak of the dot-com bubble.” However, the risks have not completely disappeared: concentration remains significant, and the market is still heavily dependent on a small number of leaders, such as Amazon, Microsoft, and Alphabet. Nevertheless, the data suggests that the current AI cycle rests on much more solid foundations than a mere speculative surge.
Adoption is very real and is becoming widespread.
The second pillar of Corinne de Boursetty’s argument is adoption. Artificial intelligence evolves in successive phases, from generative AI to more agentic and physical applications, and its main functions are automation, innovation, and productivity. What matters to investors, therefore, is that adoption is no longer limited to infrastructure providers or a handful of platforms. Today, sectors such as finance, telecommunications, healthcare, and retail are finding ways to create value using AI.
The data confirms this. Corinne de Boursetty explains that ChatGPT reached approximately 800 million weekly active users by February 2026 and that “we are seeing particularly rapid adoption” among white-collar workers in sectors such as information technology, scientific services, education, and finance. This observation suggests that the AI cycle is not limited to chips and cloud infrastructure but also encompasses integration within companies and productivity gains across the entire real economy.
Capital expenditure remains solid, but financing is more important.
The next challenge is investment. Indeed, one of the major questions facing the market concerns the ability of hyperscalers to maintain the current pace of capital expenditures. On this point, Corinne de Boursetty was clear: hyperscalers’ spending on AI amounted to USD 455 billion in 2025, and the consensus estimates that it will reach USD 712 billion in 2026 and USD 834 billion in 2027, suggesting a very large-scale infrastructure deployment, even if growth is likely to slow over time.
Corinne de Boursetty also emphasized that, as the share of capital expenditure in operating cash flow approaches 97%, the quality of financing is becoming a key priority. This does not mean the model is flawed, but it should encourage investors to scrutinize disciplined balance sheet management more closely.
Data from the AI supply chain also remains positive, with an upward cycle in the memory sector, particularly high-bandwidth memory (HBM), indicating that AI demand continues to face real capacity constraints rather than a surplus. The spending cycle therefore appears to be responding to genuine shortages and is not simply a phase of overspending.
“Valuations in the Chinese technology sector are attractive” thanks to these strengths: an advantage in valuation, competitiveness and innovation, superior profit growth, and the adoption and monetization of AI.
“Valuations in the Chinese technology sector are attractive.”
Jian Bo Gan then addressed Lombard Odier’s positioning in Asia, stating that “valuations in the Chinese technology sector are attractive” thanks to these strengths: valuation advantage, competitiveness and innovation, superior earnings growth and the adoption and monetization of AI.
The comparison of earnings between China and the United States is particularly striking. Jian Bo Gan explains that, for 2026, the expected EPS growth was 39% for the MSCI China index, compared to 27% for the MSCI USA index in the information technology sector, 34% versus 31% in the communication services sector, and 29% versus 9% in the consumer discretionary sector. Yet, the Chinese communication services and consumer discretionary sectors still have valuations significantly lower than those of their American counterparts.
China is also leading the way in adoption and implementation. As Jian Bo Gan pointed out, in 2023, China surpassed Germany, Japan, and the United States in the number of industrial robots per 10,000 workers. Furthermore, the growth of installed capacity in Chinese data centers is expected to continue. According to him, “China’s true strength lies in its application of cutting-edge technologies to daily operations.” Moreover, “faster adoption and commercialization would also accelerate monetization opportunities.” Given that the AI lifecycle is increasingly constrained by both computing power and electricity consumption, energy availability and rapid deployment are becoming strategic priorities in China.
The software sector after their correction
Market fears are currently most pronounced in the software sector, with investors worried that, in some cases, existing products could be replaced by AI agents. However, Jian Bo Gan takes a more nuanced stance: “Software related to cybersecurity, critical infrastructure, and data infrastructure” is less exposed to risk, he stated, while “core applications and content” are more vulnerable, suggesting that investors are not entirely avoiding the software sector, but are exercising discernment.
This distinction is crucial, as the software sector has already experienced a significant downturn. As Corinne de Boursetty points out, “AI deployment and adoption are driven by software.” The conclusion for investors is equally clear: the MSCI index has offered attractive opportunities since the sharp corrections it has undergone. In other words, the market may have lacked discernment in assessing the risk of disruption, a shortcoming that reveals a more appealing entry point among quality software vendors, dependent on enterprise adoption.
Implications for investment
When asked if profits could really be maintained in the event of a slowdown in the growth of investment spending in AI, Corinne de Boursetty acknowledges that the latter could not indefinitely show annual growth of 60%.
Regarding China, Jian Bo Gan concedes that regulation remains an important factor, suggesting that the current climate is significantly more accommodating than at the time of previous repressive measures since technology is now seen as a strategic priority.
For investors, the overall outlook remains optimistic, provided they are selective. In developed markets, the focus is on “major beneficiaries of AI across the value chain,” with a preference for enablers such as advanced semiconductors and new users such as computing platforms, enterprise software publishers, and cybersecurity firms. In Asia, the priority is on new users and AI enablers in China, Korea, and Taiwan.
Investors should pay attention to valuations, as well as concentration and financing risks, while exercising greater discernment in the software sector and in regional exposure.
Rethinking a heated and polarized debate
John Woods, Corinne de Boursetty and Jian Bo Gand have demonstrated that investing in AI cannot be reduced to a binary choice: to decide how to invest, it is necessary to break down the cycle into its investment components such as valuations, profits, adoption, sectoral differences or even regional opportunities.
Viewing the AI cycle from this perspective, we see that it rests on solid foundations: strong earnings growth, rapid adoption, and real demand for infrastructure. At the same time, investors must pay close attention to valuations, as well as concentration and financing risks, while exercising greater discernment in the software sector and in regional exposure.
At Lombard Odier, our role is to offer both perspective and conviction. Faced with the noise and polarization that the AI sector generates, it is more necessary than ever to step back and rethink the dominant narrative in order to identify the most attractive opportunities in terms of risk and return.
