Sound Advice: September 19, 2026

Is AI good news or bad news for investments?

AI is potentially very good news for investors—but not automatically good news for every investment. The key distinction is between AI as an economic technology and AI stocks at today's prices.

As of September 2026, there are compelling reasons for both optimism and caution:

• The bullish case: AI spending remains enormous. Gartner estimates global AI spending at $2.7 trillion in 2026, up 49.5% year over year. If AI produces sustained productivity gains, corporate profits could expand well beyond the companies currently selling AI hardware.
• The valuation case: Investors have already priced in a lot of success. Valuations of major AI companies are elevated and expected long-term earnings growth is well above historical norms.
• The spending-risk case: The five largest hyperscalers are expected to spend more than $1 trillion on AI-related capital expenditure across 2025–26. If the resulting revenues and productivity gains don't justify that investment, profits and stock valuations could suffer.
• The financing risk: Increasingly, AI infrastructure investment is being financed with debt and private credit rather than just company cash flow. That can amplify the consequences if AI returns disappoint.
• The opportunity is broader than AI companies: AI can benefit semiconductor suppliers, data centers, networking, electricity generation/transmission, cooling, cybersecurity, software, and eventually companies that use AI to reduce costs. The IMF emphasizes potentially large productivity effects, although the magnitude and timing remain uncertain.

Even so, although AI technology will probably be economically transformative, there’s no guarantee that AI stocks will outperform.

That's the same lesson from previous technological revolutions: A technology can dramatically improve the economy while some investors still overpay for the companies associated with it.

Right now, the market is particularly sensitive to evidence that AI spending will or won't generate sufficient returns. Reuters reported that investors have become nervous about a potential slowdown in AI spending, while other analysts argue that underlying demand remains strong.

AI is neither inherently good nor bad. It’s a powerful technology whose effects depend heavily on how people build, deploy, and regulate it.

The good

• Medicine: faster drug discovery, better diagnostics, and assistance for clinicians.
• Productivity: automates repetitive work and helps people create, analyze, and learn faster.
• Education: personalized tutoring and broader access to expertise.
• Science: helps researchers process enormous datasets and explore possibilities faster.
• Accessibility: translation, speech assistance, and tools for people with disabilities.

The risks

• Jobs: some occupations and tasks may be displaced or substantially changed.
• Misinformation: AI makes convincing fake text, images, audio, and video much cheaper to produce.
• Privacy: AI systems can enable large-scale collection and analysis of personal information.
• Concentration of power: advanced AI could give disproportionate advantages to a relatively small number of companies or governments.
• Security: AI can make certain cyberattacks and other harmful activities easier.
• Errors and bias: AI can confidently produce incorrect or unfair results.

The bigger picture

AI is best viewed as a force multiplier. It can amplify human capabilities—and, when poorly designed or misused, amplify human mistakes and harmful intentions.

The most consequential question probably isn't "Is AI good or bad?" but "Who controls it, what incentives govern its use, and what safeguards accompany its deployment?"

“How much AI success does the stock market already assume?”, the answer is: a substantial amount—especially in the biggest AI-related companies—but probably not all of the potential economic benefit.

As of September 2026, I'd separate it into three layers:

1. The obvious AI winners: a lot is already priced in

The market is already assigning very high expectations to companies at the center of the AI buildout. The Bank for International Settlements says the largest AI companies have implied long-term earnings growth well above historical benchmarks.

At the same time, valuations have actually come down from their 2025 extremes. RBC reports that a broad basket of AI/technology companies is around 20× forward earnings, versus 28.5× in October 2025.

So this isn't simply “AI stocks are wildly expensive.” It's more nuanced:

The market expects extraordinary growth, but investors have already adjusted prices for some of the risks.

2. The broader stock market: Less AI is priced in than you might think

This is where it gets interesting.

Goldman Sachs estimates that AI infrastructure beneficiaries could account for roughly half of S&P 500 earnings growth in 2026–27. Its forecasts also incorporate only about a 0.4 percentage-point boost to S&P 500 earnings growth from AI productivity in 2026 and 1.5 points in 2027.

That suggests the market is currently pricing a lot of AI infrastructure spending, but perhaps considerably less of the eventual productivity revolution.

In other words:

AI chips and data centers are already heavily anticipated. AI-driven transformation of the rest of the economy may be less fully reflected in prices.

That's an important distinction.

3. The enormous unknown: the return on all that spending

This is probably the single biggest investment question.

The five largest hyperscalers are expected to spend more than $1 trillion on AI-related capex over 2025–26. BIS warns that this investment is running ahead of the companies' earnings and free cash flow, creating the possibility that some AI investments won't generate adequate returns.

There's encouraging evidence too: cloud revenues and margins are improving, and JPMorgan says AI capex is already contributing to earnings growth. But it also notes that only 2% of S&P 500 companies that mentioned AI in recent earnings calls quantified an AI-driven productivity impact on profits.

That's the missing piece.

Here’s a useful way to think about it

Imagine three possible futures:
AI disappoints	Current prices could prove too high
AI works roughly as expected	A lot of the benefit may already be reflected in prices
AI dramatically boosts productivity	Some future benefits may still be underappreciated

The third scenario is why I wouldn't conclude that “AI is already priced in, therefore avoid AI.”

But the first two explain why being right about AI doesn't automatically mean making money from AI stocks.

There's another important point: the S&P 500 itself is now unusually concentrated in technology. Reuters reported this week that information technology represented about 38% of the index, versus roughly 18% historically. The index's forward P/E was about 20.3, above its long-term average but substantially below the extremes reached during the dot-com bubble.

My bottom line

I'd frame the situation this way:

The market has already priced in a significant AI boom. It has not necessarily priced in the full economic consequences of AI.

That means the investment opportunity isn't necessarily “buy anything related to AI.” The more interesting question is which companies can turn AI spending into durable cash flows, and which companies' current prices require near-perfect execution.

N. Russell Wayne

Weston, CT  06883

203-895-8877

www.soundasset.blogspot.com