Goldman Sachs has warned that AI’s potential to disrupt businesses is starting to reshape how investors value US stocks, with the biggest concern landing squarely on the software sector and other high-growth names. For South African investors watching global markets, the message is clear: the artificial intelligence boom is no longer just about excitement and hype, but about whether today’s share prices are being stretched by very optimistic assumptions about future earnings.
The Wall Street bank said the market is leaning heavily on what it calls terminal value — the worth of a company beyond the next decade, based on expected earnings growth. According to Goldman, that portion now makes up about 75% of the S&P 500’s equity value, which is close to a 25-year high. In plain language, that means a huge chunk of today’s market value depends on profits that may only arrive many years from now.
Goldman’s warning matters because it echoes a familiar theme from the dot-com era: when investors believe a technology shift will rewrite the economy, they often pay up for growth long before the numbers justify it. The bank said this latest surge in long-term expectations is “elevated versus history” and resembles previous periods of market exuberance, including the dot-com boom.
For readers in South Africa, the lesson is not limited to Wall Street charts. When global tech stocks wobble, the effects can spill into pension funds, local asset managers and the broader risk appetite that influences markets from Johannesburg to Cape Town. A sharp reset in US tech can quickly filter through to emerging markets, especially when global money starts questioning whether AI will pay off as fast as companies promise.
Investor nerves have been building for months, especially after Anthropic rolled out new tools that can automate tasks across areas like marketing and data analytics. That has raised a tougher question for the software industry: if AI can do more of the work, what happens to revenue growth, pricing power and margins for traditional software firms? It is exactly the sort of question that can make a market nervous when valuations are already high.
The S&P 500 software and services index has fallen by about 17% this year, with much of the weakness driven by fears that AI tools could erode future earnings. The concern is not necessarily that these companies will disappear, but that the profit models investors once relied on may need to be rewritten.
Big Tech, meanwhile, is still spending aggressively to stay ahead. Alphabet, Microsoft, Meta and Amazon have committed billions of dollars to AI-related capital expenditure over the next three years, locked in what is becoming a fierce race for dominance in the next phase of computing. Yet even with those heavy commitments, investors continue to ask the same blunt question: when does the return show up?
Goldman Sachs says the AI bubble debate is far from over
Goldman’s latest note suggests the AI bubble debate will remain a live issue for several quarters, mainly because markets are still unsure how much of current valuation is justified by real profits versus future hope. The bank said the big four cloud companies are on track to spend about US$600-billion on AI infrastructure this year alone, an extraordinary level of investment that has already tightened cash flows and tested investor patience.
That scale of spending tells its own story. The companies leading the AI race are effectively betting that the infrastructure they build now will create enormous revenue streams later. But the market has become less forgiving of long waits, especially after years of rising interest rates made investors more sensitive to cash flow, balance sheet strength and actual delivery.
Goldman also put some numbers behind the risk. It estimated that every 1 percentage point decline in assumed long-term growth would reduce the combined enterprise value of S&P 500 companies by roughly 15%. High-growth stocks would be hit much harder, with valuations dropping by about 29%, compared with around 10% for low-growth equities.
That gap helps explain why investors are so fixated on which businesses may benefit from AI and which ones may be exposed. If a company’s value is built mainly on earnings far in the future, even a modest change in growth assumptions can wipe out a large slice of market capitalisation. Goldman said this is why the value of a high-growth company is “especially sensitive” to shifts in its long-term outlook.
The bank added that uncertainty around disruption is likely to hang over markets “until later stages of AI adoption”. In other words, the debate is not going away anytime soon. As more companies deploy AI in customer service, coding, sales, analytics and back-office operations, the pressure on existing software models will only intensify.
One striking detail from Goldman’s review of recent earnings calls is how short-term-focused companies still are. The bank said only 5% of S&P 500 companies discussed financial metrics beyond five years in recent quarterly calls. That, it argues, leaves investors with too little guidance on where management teams think their businesses are really heading.
Goldman said more executives should be willing to speak openly about the long-term picture, especially as AI begins to change everything from productivity assumptions to pricing strategy. For investors, that kind of transparency could help separate real transformation from marketing spin.
The broader market takeaway is that AI remains powerful, but it is also forcing a reset in how value is measured. For now, the winners are the firms with the scale and cash to keep spending. The losers could be those whose business models look vulnerable to automation.
As we reported earlier, the tension in global markets is no longer about whether AI is important — it clearly is. The bigger question is whether current prices have already assumed too much, too soon. Goldman Sachs clearly thinks that debate is just getting started, and for South African investors with global exposure, that is a warning worth keeping close.