Goldman Sachs sees new inflation pressure from AI

It's safe to say that inflation in the U.S. this year felt a lot like a game of whack-a-mole. Just as one pressure point starts to ease, another pops up somewhere else.

The turnaround many had hoped for never arrived. Instead, the Fed's preferred inflation gauge, PCE, rose to 4.1% in May from 2.9% in February, according to CBS News, while core PCE remained stubbornly above the Fed's target.

However, the market's primary catalyst, artificial intelligence, was supposed to swoop in and save the day like Superman.

Many, especially those reading Nvidia reports like gospel, expected the AI boom to make the economy faster, smarter, and cheaper.

Instead, according to a Business Insider report, Goldman Sachs warns it may first show up as another inflation shock, with Americans helping pay the bill through pricier software, power, and tech hardware.

Why Goldman Sachs sees an AI inflation problem for Americans

Goldman Sachs warns that the relentless AI buildout will come at a much higher cost to Americans in terms of inflation.

What the AI buildout has done is stoke demand for memory chips, software, and electricity, among other things, amid supply constraints.

In Goldman's view, the pressure is showing up in the Fed's preferred inflation gauge.

Megan Peters, an economist at the bank, estimates AI is lifting U.S. core PCE inflation by about 20 basis points a year. By year-end, that drag could more than double, with the boost to core PCE reaching 50 basis points.

The impact is far greater than the likely effect in Canada, Australia, Europe, the U.K., and Japan, where Goldman sees an average 10-basis-point increase.

"While not completely negligible, these effects are far below the 50bp peak we estimate for U.S. PCE, suggesting that for the most part, AI-driven inflation is a U.S. story," Peters wrote.

The pressure stems from three major factors: soaring memory and software prices, and rising electricity demand.

How memory-chip prices are feeding the inflation warning

The first pressure point is memory, something iPhone fans are probably tired of hearing about, as are video gaming fans like me.

"Unfortunately, price increases are unavoidable," CEO Tim Cook said in an exclusive Wall Street Journal interview published on June 17, 2026.

More AI:

• The new Chinese AI model rattling U.S. tech investors
• Anthropic restores access to Mythos 5 for select organizations
• SoftBank CEO offers stinging critique of Musk's AI bet

Goldman Sachs concurs with Cook that the tremendous demand for AI hardware is driving up the cost of key components, which flow into consumer technology, business software, and broader digital services.

For example, Keepa data shows Corsair Vengeance DDR5 RAM 32GB on Amazon surged from about $110 to $415 over the past year, a nearly 277% increase.

In a Bank of America note shared with me, analysts also say that memory now represents roughly 35% to 40% of cloud AI capex, two to three times the historical level.

They also doubled down on Micron stock, assigning a $1,550 price target.

That said, Goldman expects the pressure to peak before the end of 2026. Prices in that category could be rising at a 30% year-over-year pace by November.

The U.S. is more exposed because software and accessories account for about 1% of PCE inflation, compared to less than 0.5% in other developed economies.

Why software may become the next AI price shock

The second big inflationary shock comes from software, where AI is being layered into products that businesses and households already use.

Once AI tools are a pivotal part of the whole package, the cost can be passed through in subscription prices rather than shown as a separate add-on.

Taking Microsoft as a clear example, the tech giant recently moved Copilot deeper into Microsoft 365, while commercial suite prices are rising, including Office 365 E3 to $26 from $23 and Microsoft 365 E3 to $39 from $36 starting July 1, 2026, Agolution reported.

Google made a similar move with Workspace, announcing that Gemini AI is now included in the Business and Enterprise plans.

How data centers strain electricity costs

Goldman Sachs' third big inflation wave is electricity, and it's probably the one that's hardest to ignore.

Naturally, AI needs a ton of data centers, cooling systems, and chips, as well as a power grid that's capable of handling a much heavier load.

The report identified that the average price of one kilowatt-hour of electricity in a U.S. city rose to $0.19 in May, up about 27% from May 2022, according to the Bureau of Labor Statistics.

That adds another painful bill to an already uncomfortable stack.

Goldman estimates data centers could account for about 11% of total U.S. power demand by the end of the decade, roughly double today's roughly 6% share.

Though it's a huge challenge on its own, the AI buildout is happening at a time when energy markets are already tense, with oil prices still up sharply year-to-date amid geopolitical supply fears.

The fragile ceasefire between the United States and Iran has now collapsed, reviving fears of another spike in oil prices after prices had declined over the past several weeks.

However, over the long term, investors are latching onto the idea that AI will be disinflationary and boost productivity.

Cathie Wood talked about it during ARK Invest's January "In The Know" webcast. "Productivity-driven growth is associated with falling inflation," she said.

Nevertheless, Goldman's warning is that the bill may come first, while the benefits arrive later.

Related: Bernstein revamps gold price target on Fed-rate shift

The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.

This story was originally published July 12, 2026 at 1:17 PM.