Morgan Stanley Lifts 2026 US GDP Forecast to 2.3% on Inelastic AI Demand

A new report from Morgan Stanley suggests the global economy is being reshaped by "inelastic demand" for artificial intelligence infrastructure, prompting the bank on May 26 to raise its 2026 US GDP growth forecast to 2.3% on the back of a capital spending boom that it says is ignoring rising costs.

"For the market, the key question may no longer be 'has the price gotten high enough to affect demand?' but rather 'has demand become too strategic, too necessary, or too well-funded to care about price?'," the bank's analysts wrote in their mid-year outlook.

The bank's forecast for AI-related capital expenditure by US hyperscalers in 2026 was nearly doubled to $805 billion from a projection of $433 billion made just a year ago. The report sees spending climbing to $1.1 trillion in 2027. This surge prompted Morgan Stanley to lift its 2026 S&P 500 earnings growth forecast from 17 percent to 23 percent and its US business fixed investment growth outlook from 3 percent to 7 percent.

The report argues this inelastic spending creates a durable floor for the economy, but it also introduces significant concentration risk. With the 10-year Treasury yield near 5 percent, the market's reliance on a handful of AI-linked stocks for its returns poses a critical question for investors: is the AI boom a broad-based productivity miracle or a narrow construction boom with systemic risk?

The Inelasticity Engine

Morgan Stanley's central thesis is that AI investment has become a strategic necessity for tech giants, who fear being left behind in the next technological wave. This has created a demand curve that is unresponsive to traditional economic pressures. Hyperscale firms are pouring money into data centers, chips, and power infrastructure at a pace that doesn’t slow when input costs for components like memory chips or copper rise, or when financing becomes more expensive. This dynamic, usually reserved for essential goods like electricity, is now driving an unprecedented physical buildout.

A Market of One

This concentration of spending has led to a similar concentration in stock market returns, creating a potential blind spot for passive investors. Analysis of market data shows that while the SPDR S&P 500 ETF (SPY) returned 41% over the two years ending in May 2026, the return would have been a far more modest 16% if AI infrastructure companies like Nvidia and AMD were excluded. According to one report, Nvidia alone now accounts for 8% of the entire S&P 500 index by weight. This heavy concentration in a few highly-valued stocks, such as AMD with a forward P/E of 67, exposes index investors to significant single-sector risk.

GDP: Headline vs. Reality

While Morgan Stanley's report paints a bullish picture for US growth, research from Goldman Sachs offers a more tempered view. After accounting for the fact that much of the AI hardware is imported, Goldman's analysis from 2025 suggested AI’s direct contribution to US growth was only about 0.2 percentage points of the total 2.2% growth recorded that year. A large portion of the billions being spent on the AI buildout is flowing to overseas chip manufacturers like TSMC and equipment suppliers, thinning the direct domestic economic benefit.

Investment Implications

Morgan Stanley's conclusion is to overweight US equities, citing the inelastic demand as a powerful domestic growth engine. The bank also suggests going long on oil as a hedge, assuming the strong economic activity will support energy demand. However, the data on market concentration offers a sharp counterpoint. For retirement portfolios and passive funds, the outsized gains of the past two years have been driven by a handful of AI-related megacaps. The risk now is that these portfolios are far more exposed to the fortunes of a single sector than a diversified index would suggest.

This article is for informational purposes only and does not constitute investment advice.