The AI trade is changing, not disappearing
Corrado Tiralongo - Aug 20, 2026
In his latest article, Corrado Tiralongo examines how the AI investment theme is evolving as markets become more selective.
Semiconductor stocks have fallen sharply, but the broader equity market has largely held its ground.
The weakness has been concentrated among the companies that benefited most directly from enthusiasm surrounding artificial intelligence. Other parts of the market have performed considerably better. The equal-weighted S&P 500 recently reached a record high, while software companies and several non-technology sectors have helped offset the decline in chipmakers.
This is not yet a broad market breakdown. Investors are still allocating capital to equities, but they have become more selective about where the benefits of AI will accrue and how much they are willing to pay for them.
The AI investment theme is not disappearing. It is entering a more demanding phase.
Why semiconductor stocks have come under pressure
Several concerns have converged on the semiconductor industry.
Chinese memory-chip manufacturers are expanding their capabilities, raising the possibility of greater competition and lower prices. China remains well behind the global leaders in high-bandwidth memory and the most advanced chips required for AI. Greater production of mainstream memory chips could nevertheless alleviate current shortages and place pressure on industry revenues.
China is also making progress in semiconductor manufacturing equipment. Its reported production of domestic deep ultraviolet lithography machines remains well behind the most advanced technology available from Western suppliers, but it points to a gradual reduction in China’s reliance on foreign technology.
The economics of AI development are also attracting greater scrutiny. Chinese developers have reported training near-frontier models at substantially lower costs than their American counterparts. Comparisons are difficult, and the cost advantage in running these models is less clear. Even so, the results raise questions about whether the enormous capital commitments being made by US technology companies will generate the returns investors currently expect.
Financing within the AI ecosystem has added another layer of uncertainty. Potential financing arrangements between chip suppliers and their customers make it more difficult to determine where underlying demand ends and financing-supported demand begins.
AI’s long-term potential remains substantial, but the assumptions embedded in semiconductor valuations are facing a more rigorous test.
When rotation becomes a broader correction
For now, the decline looks more like a rotation within the equity market. Investors have moved capital from semiconductor producers toward software companies, AI users and other potential beneficiaries.
A wider base of market leadership is generally constructive after several years in which returns were dominated by a small number of technology companies. It does not, however, guarantee that the rest of the market will remain insulated if technology weakness intensifies.
There is a historical warning here. When technology stocks began falling in 2000, the average US stock initially performed considerably better. That rotation delayed the effect on the headline index, but it did not prevent the technology correction from eventually weighing on the market as a whole.
The current environment is not a replay of 2000. Corporate balance sheets, business models and the underlying demand for AI are very different. The comparison is still useful because it shows that broadening and rising market risk can exist at the same time.
The dividing line is the source of the weakness. A semiconductor correction caused by greater supply and lower expected profitability can remain largely a rotation. A correction driven by weakening AI demand would carry much greater consequences. Slower corporate adoption, reduced capital spending by hyperscalers or disappointing productivity gains would challenge earnings assumptions across the technology sector.
Earnings raise the bar
The latest earnings results suggest that AI demand remains strong. They also show that investors have become far less forgiving.
SK Hynix reported record quarterly revenue and operating profit, supported by strong demand for AI memory. Its operating profit increased more than sixfold from a year earlier. Yet the results fell short of elevated forecasts, and its shares declined 9.6%. Extraordinary growth was not enough because investors had expected even more.
Microsoft produced the opposite reaction. Azure revenue grew 43%, while the company provided stronger evidence that its AI and cloud investments are translating into revenue and cash generation. Its shares rose 15.5% following the results.
Meta and Amazon reinforced both sides of the debate. Meta’s revenue increased 28%, but operating income declined and quarterly free cash flow fell sharply as costs and capital investment rose. Amazon reported 37% growth from Amazon Web Services, its fastest rate in more than four years, but trailing 12-month free cash flow turned negative as infrastructure spending accelerated.
Demand is not the immediate problem. The question is whether the escalating cost of meeting that demand will ultimately produce adequate returns.
A near-term technology rebound would not resolve that uncertainty. Valuations have declined, creating room for a recovery if earnings remain strong and sentiment improves. A more difficult test would come if analysts begin reducing earnings forecasts as they reassess the returns likely to be generated by today’s capital spending. A recovery followed by a larger correction is therefore entirely plausible.
The market is shifting from asking who is spending on AI to asking who is earning from it.
How the Canada Life Investment Management portfolios are positioned
Our portfolio construction does not depend on the continued leadership of a single investment theme.
We maintain exposure to AI and believe it will remain an important driver of innovation, productivity and corporate investment. Our participation is diversified across companies, industries and regions. This includes businesses supplying AI infrastructure as well as companies that may use the technology to improve productivity, strengthen margins or develop new sources of revenue.
We emphasize capital discipline and the sustainability of free cash flow. When expectations are already elevated, the ability to convert investment into durable earnings carries more weight than the size of an AI announcement or capital-spending budget.
Beyond technology, our portfolios maintain exposure to fixed income, liquid alternatives, private assets and a broader range of equity return drivers. These allocations are intended to limit dependence on any single market narrative and diversify the sources of return and risk within the portfolios.
AI may prove to be one of the defining investment themes of this decade. That does not mean every participant will succeed or that investors will earn attractive returns at any price.
The AI train has not come off the tracks. Investors are becoming less willing to pay for the journey without clearer evidence of where it leads and who will profit along the way.
Corrado Tiralongo (he/him)
Vice President, Asset Allocation & Chief Investment Officer
Canada Life Investment Management Ltd.