Three years after ChatGPT kicked off the AI frenzy, Wall Street’s tone is shifting. The money is still flowing into artificial intelligence, but so are doubts about whether the returns will justify the spending. Recent selloffs in Nvidia $NVDA ( ▲ 3.74% ) and Oracle $ORCL ( ▲ 7.58% ) have sharpened the debate over whether the AI trade is nearing a breaking point or just pausing before its next leg higher.
Signs of skepticism are stacking up. Oracle shares slid after the company revealed heavier than expected AI spending, while sentiment around companies tied to OpenAI has cooled. Looking ahead to 2026, investors are split between trimming AI exposure to avoid a bubble pop or doubling down on what could still be a generational technology shift.
When the Spending Gets Real
The unease centers on cost. Developing and scaling AI is enormously expensive, and it remains unclear how much consumers are ultimately willing to pay. That matters because the S&P 500’s massive three year rally has been driven largely by Big Tech names like Alphabet $GOOGL ( ▲ 0.72% ) and Microsoft $MSFT ( ▼ 0.05% ) , alongside AI infrastructure beneficiaries like Nvidia $NVDA ( ▲ 3.74% ) , Broadcom $AVGO ( ▲ 1.98% ) , and power providers such as Constellation Energy $CEG ( ▼ 0.35% ) . If growth slows, the broader market could feel it quickly.
Investors say these stocks do not fall because growth dips slightly. They fall when growth stops accelerating. That risk is now front and center as expectations get harder to beat.
Capital, Credit, and Cracks in the Story
Access to capital is another fault line. OpenAI alone plans to spend roughly $1.4 trillion in the coming years and is expected to burn cash until at least 2030. So far, fundraising has not been an issue, with billions raised from SoftBank and others, and even direct investment commitments from Nvidia. But some investors worry this creates circular financing, where chipmakers fund customers who then buy more chips.
Oracle offers a clear example of how this can turn. The company booked massive cloud demand but financed data center construction with tens of billions in debt. After reporting higher capex and weaker cloud growth, Oracle shares slid further on reports of delayed OpenAI related projects. Credit markets took notice too, with measures of Oracle’s credit risk hitting their highest levels since 2009.
That stress does not stay isolated. When one AI spender stumbles, it ripples through suppliers, infrastructure firms, and the broader AI ecosystem.
Big Tech’s Costly Pivot
Alphabet $GOOGL, Microsoft $MSFT ( ▼ 0.05% ) , Amazon $AMZN ( ▲ 0.87% ) , and Meta $META ( ▲ 0.32% ) are projected to spend more than $400 billion on capital expenditures over the next year, largely on data centers. While AI driven revenue is growing, it is nowhere near covering those costs yet.
Depreciation is becoming a real problem. Combined depreciation expenses for Alphabet, Microsoft, and Meta jumped from about $10 billion in late 2023 to nearly $22 billion this year and are expected to approach $30 billion next year. That pressure could squeeze buybacks and dividends, a key reason investors have long loved Big Tech.
This marks a major shift. These companies built their reputations on fast growth and low marginal costs. AI flips that model by demanding massive upfront investment with uncertain payoff.
Not Dot-Com Levels, but Not Cheap Either
Despite the anxiety, valuations are not at dot-com extremes. The Nasdaq 100 trades around 26 times forward earnings, far below the 80 plus multiples seen in 2000. Most of the Magnificent Seven are below 30 times earnings, which many investors see as reasonable given their scale and profitability.
That said, pockets of speculation exist. Palantir $PLTR and Snowflake $SNOW trade at triple digit earnings multiples, standing out even among AI favorites. Still, the biggest names driving the AI boom are not priced like classic bubbles, which is why the debate remains unresolved.
For now, Wall Street is stuck in the middle. The risks are obvious, the spending is massive, and the stakes are high. The AI trade may not end in a dramatic crash like 2000, but many investors expect a rotation once the growth narrative shows its first real cracks.