喧嚣之上：人工智能、盈利与长期看空者

AI demand may be different from past booms because the need for chips, data centers, energy, and computing power already exists and exceeds supply.

Stronger earnings, flat hiring, and broadening market leadership suggest productivity gains may already be supporting the rally.

Bearish concerns serve a purpose, but resilient growth, relatively well-anchored inflation expectations, and available liquidity have challenged the market-top narrative.

When people ask me what period today resembles, I suspect they want me to provide an historical episode associated with a market crisis. Perhaps 1873 or 1893, when excessive railroad expansion and speculative financing ended in crisis. Or maybe 1999 and 2000, when investors poured capital into internet startups long before viable business models emerged.

But my answer is likely unsatisfying. I don’t think there is a parallel to crises of the past.

Railroads and the internet were transformative technologies, but in both cases, capital raced ahead of demand. Too many rail lines were built for the traffic they carried. Too many internet companies were funded before customers, revenues, and profits existed. Technology changed the world, but much of the investment arrived before the economic reality.

Today’s environment looks different. The demand isn’t hypothetical. Demand for computing power has already exceeded supply. Companies are scrambling for chips, memory, data centers, energy, and network infrastructure because the need already exists. As artificial intelligence (AI) evolves from answering questions to deploying agents that work continuously, the appetite for computational resources should only grow. This is less a story of capacity chasing uncertain demand and more a story of demand straining against limited capacity.

If anything, it might resemble 2023. Back then, many investors worried that war, elevated oil prices1 and rising bond yields2 would eventually undermine the economy and stock markets. Sound familiar? Yet, inflation expectations had peaked in the prior year,3 corporate earnings had been strong,4 and stocks finished the year with strong gains.5

My apologies for the boring answer. Sometimes the most important thing about history is anticipating when the worst outcomes aren’t likely to be repeated.

… the much-anticipated productivity gains from AI and other technological advances may already be showing up in the data. Consider that US corporate earnings rose roughly 25% over the past year,6 while nonfarm payroll growth has been essentially flat.7 Historically, earnings growth of that magnitude has been accompanied by meaningful hiring. Today, many companies have been earning more without adding large numbers of workers. And this isn’t just a story confined to technology companies or the US. Evidence of stronger earnings has been seen across sectors and across global markets.

“I love the permabears. They do all the hard work of identifying what could go wrong, which allows me to focus on evaluating what they may be missing, and to concentrate on what could go right.”

Amen. I couldn’t have said it better myself. What have the permabears served over the past year?

Nonetheless, I’ll keep listening to the bears. Their arguments help challenge assumptions, test convictions, and guard against complacency. In that sense, they serve an important purpose. But as Ed Yardeni suggested, they’re also doing some of the negative work for the rest of us, freeing up our time to focus on where the opportunities are.

A: There’s a narrow interpretation and a broader, more important one. The narrow view is that policymakers were simply responding to a disorderly decline in the yen and attempting to stabilize currency markets.

Instead, I viewed the intervention as a signal that US policymakers have been increasingly focused on preserving stability across interconnected markets, particularly the Treasury market. With Japan a major holder of US government debt and long-term Treasury yields having moved meaningfully higher, the coordinated action and the emphasis on mechanisms that reduce the need for Treasury sales suggested to me that the US Treasury stands ready to support market functioning if necessary.

At a time when investors are concerned about rising deficits, growing debt issuance, and higher 30-year Treasury yields,13 I believe the intervention sent a message that policymakers are paying close attention to these pressures and are prepared to respond if market moves become destabilizing.

A: I’d be careful about viewing Citadel’s rescue financing of Situational Awareness as a classic market top signal. Market tops have typically been associated with deteriorating fundamentals, weakening earnings, tightening liquidity, or a broad loss of confidence. In this case, the issue appears to have been much more idiosyncratic. Situational Awareness was heavily leveraged during a period of market consolidation rather than a period in which the underlying economic or corporate earnings backdrop appeared to be unraveling. When leverage meets short-term volatility, even in an otherwise constructive environment, accidents can happen.

It’s also worth noting that the availability of capital willing to step in and provide rescue financing can be viewed as a source of strength rather than a warning sign. Distressed investors exist for moments like these, providing liquidity when individual market participants find themselves under pressure. Rather than exposing a systemic problem, I believe the episode demonstrated that there’s capital available to absorb isolated stress. The market’s reaction reinforced that point in my view. Since the period in which Situational Awareness encountered difficulties, the US tech sector has already retraced much of its prior decline.14

Think: Token volumes aren’t growing fast enough to justify the AI infrastructure build.

Rethink: Agentic workflows could dramatically increase AI usage because they do far more than answer a single prompt. A typical human interaction may consume only a few thousand tokens, while an AI agent that plans, researches, analyzes, drafts, critiques, and revises can consume hundreds of thousands or even millions of tokens.15

US Treasury rates have risen in recent weeks. I reached out to Rob Waldner, Head of Invesco Fixed Income, to hear his views. Here’s his response:

“We maintain a neutral stance on US interest rates. The bond market has priced in almost two Fed rate hikes over the next year,16 although we expect policy rates to remain on hold, because of easing inflation pressures17 and trend-like economic growth.18 Under normal circumstances, current market pricing could provide sufficient risk premium to justify a long duration position. However, elevated geopolitical risks and uncertainty may argue against taking a more directional view. With both uncertain economic and geopolitical outcomes, we believe the balance of risks has been broadly even, and therefore we remain neutral on longer-term rates.”

My travels recently took me to Hong Kong and Beijing. The world doesn’t need another market strategist pointing out China’s advances in high-tech exports, open-source AI, electric vehicles, robotics, and countless other industries. What stood out to me was something far less quantifiable. I’ve often said that despite the constant focus on division and conflict in the US, my experience traveling the country is overwhelmingly positive. I meet thoughtful, intelligent, and empathetic people everywhere I go. After this trip, I can say the same about China. Beneath the headlines and geopolitical competitions, I see people who, like most of us, are simply looking to enjoy life and build better futures for themselves and their families. I’m already looking forward to my next visit to a Beijing hutong bar.

Source: Bloomberg L.P., based on West Texas Intermediate crude sweet oil.

Source: Bloomberg L.P., based on the 10-year US Treasury rate.

Source: Bloomberg L.P., based on the 5-year US Treasury inflation breakeven, which peaked on March 25, 2022.

Source: Bloomberg L.P., based on the operating earnings of the companies in the S&P 500 Index.

Source: Bloomberg L.P., based on the S&P 500 Index, which advanced by 26.26% in 2023.

Source: Bloomberg L.P., June 30, 2026, based on the operating earnings of the companies in the S&P 500 Index.

Source: US Bureau of Labor Statistics, July 2026.

Source: Bloomberg L.P., Aug. 18, 2026, based on the price-to-earnings ratio of the S&P 500 Index.

Source: Bloomberg L.P., Aug. 17, 2026, based on the year-to-date total return of the S&P 500 Index (13.91%) compared to the S&P 500 Equal Weight Index (16.31%).

Source: Bloomberg L.P., based on West Texas Intermediate crude sweet oil.

Source: Bloomberg L.P., based on the 5-year US Treasury inflation breakeven.

Source: Conference Board, July 31, 2026, based on the Conference Board US Leading Index Ten Economic Indicators.

Source: Bloomberg L.P., Aug. 17, 2026, based on the 30-year US Treasury rate.

Source: Bloomberg L.P., Aug. 17, 2026, based on the S&P 500 Information Technology Sector, which declined 13.78% peak-to-trough from June 2, 2026 to July 29, 2026. The sector advanced 12.47% from the market close on July 29, 2026 to August 17, 2026.

Source: Goldman Sachs, “AI Agents Forecast to Boost Tech Cash Flow as Usage Soars,” May 20, 2026.

Source: Bloomberg L.P., Aug. 17, 2026, based on fed funds implied rates.

Source: US Bureau of Labor Statistics, July 31, 2026, based on the US Consumer Price Index (CPI).

Source: US Bureau of Economic Analysis, June 2026.

All investing involves risk, including the risk of loss.

Past performance does not guarantee future results.

Investments cannot be made directly in an index.

This does not constitute a recommendation of any investment strategy or product for a particular investor. Investors should consult a financial professional before making any investment decisions.

Artificial intelligence (AI) technology companies are sensitive to specific risks such as small markets, business cycle changes, economic growth, technological progress, obsolescence, and regulation. These companies may have limited products, markets, resources, or personnel, making their securities more volatile, especially for smaller start-ups. Rapid technological changes can adversely affect their results. AI companies often rely on patents, copyrights, trademarks, and trade secrets to protect their technology, but there's no guarantee these protections will be sufficient. Significant research and development (R&D) spending doesn’t ensure product or service success.

Breakeven inflation is the difference in yield between a nominal Treasury security and a Treasury Inflation-Protected Security of the same maturity.

The Consumer Price Index (CPI) measures the change in consumer prices and is a commonly cited measure of inflation.

Duration is a measure of the sensitivity of the price (the value of principal) of a fixed income investment to a change in interest rates. Duration is expressed as a number of years.

The fed funds implied rate is the difference between the spot rate and the futures rate, which is an interest rate that can be calculated for any security with a futures contract.

Fixed income investments are subject to the credit risk of the issuer and the effects of changing interest rates. Interest rate risk refers to the risk that bond prices generally fall as interest rates rise and vice versa. An issuer may be unable to meet interest and/or principal payments, thereby causing its instruments to decrease in value and lowering the issuer’s credit rating.

In general, stock values fluctuate, sometimes widely, in response to activities specific to the company as well as general market, economic, and political conditions.

Inflation is the rate at which the general price level for goods and services is increasing.

Investments focused on a particular industry or sector are subject to greater risk and can be more impacted by market volatility than more diversified investments.

Leverage measures a company’s total debt relative to the company’s book value.

Many products and services offered in technology-related industries are subject to rapid obsolescence, which may lower the value of the issuers.

Option-adjusted spread (OAS) is the yield spread that must be added to a benchmark yield curve to discount a security’s payments to match its market price, using a dynamic pricing model that accounts for embedded options.

The price-to-earnings (P/E) ratio measures a stock’s valuation by dividing its share price by its earnings per share.

The S&P 500® Equal Weight Index is the equally weighted version of the S&P 500® Index.

The S&P 500® Index is an unmanaged index considered representative of the US stock market.

Treasury Inflation-Protected Securities (TIPS) are US Treasury securities that are indexed to inflation.

West Texas Intermediate (WTI) is a type of light, sweet crude oil that comes from the US.

The opinions referenced above are those of the author as of Aug. 21, 2026. These comments should not be construed as recommendations, but as an illustration of broader themes. Forward-looking statements are not guarantees of future results. They involve risks, uncertainties, and assumptions; there can be no assurance that actual results will not differ materially from expectations.

This link takes you to a site not affiliated with Invesco. The site is for informational purposes only. Invesco does not guarantee nor take any responsibility for any of the content.