AI stocks today: Wall Street sees a split in the next wave

The AI trade is no longer moving as one block. While the Magnificent Seven have been flat as a group in 2026, semiconductor shares and smaller infrastructure companies have produced sharply different results, pushing investors to rethink where the next gains may come from. The debate now centers on whether money rotates back into hyperscalers or toward the physical bottlenecks behind the AI build-out.

What We Know So Far

The broad market picture is unusually divided. Business Insider reported that the Roundhill Magnificent Seven ETF was flat since January even as the S&P 500 climbed by double digits and semiconductor stocks surged. The group includes Nvidia, Alphabet, Meta, Microsoft, Tesla, Apple and Amazon, all closely linked to the AI investment cycle.

Morgan Stanley sees that gap as difficult to sustain. Chief U.S. equity strategist Mike Wilson wrote that the performance split between hyperscalers and semiconductor stocks was likely unsustainable because chip demand ultimately depends on spending by the large cloud companies. Goldman Sachs has also been discussing more exposure to hyperscalers, with strategist Ben Snider pointing to price-to-earnings ratios similar to levels seen around major market lows in March 2020 and October 2022, according to Business Insider's market report.

Other investors are looking beyond the biggest names. Benzinga described a shift toward companies solving physical constraints in the AI infrastructure chain: advanced chip packaging, faster optical data connections and direct-to-chip cooling. Its report said annualized generative AI revenue had passed $175 billion, while hyperscalers had committed about $2 trillion in cumulative capital spending.

The company-level numbers help explain why the debate has intensified. The Motley Fool reported that the Nasdaq CTA Artificial Intelligence Index had gained more than 60% over the previous year, compared with a 25% rise for the Nasdaq Composite. Sandisk's data center sales reached $1.46 billion in its fiscal third quarter, up 645% from a year earlier, while Palantir's first-quarter commercial revenue rose 133% to $595 million.

Hyperscaler

A very large cloud and technology company capable of spending heavily on data centers and AI computing capacity.

Advanced packaging

Technology used to combine or stack chips so they can move data faster and operate efficiently at greater scale.

Optical interconnect

A high-speed data connection that uses light rather than traditional copper links inside large computing systems.

The Response

Wall Street is not speaking with one voice, but the bullish arguments share a common theme: weaker stock prices do not necessarily mean weaker AI demand. Morgan Stanley expects investors to reward more disciplined hyperscaler spending, while Goldman Sachs sees unusually low valuations relative to previous market turning points.

The demand is broadening across industries.

Nebius is one example of the infrastructure case. The Motley Fool reported that Nvidia invested $2 billion to help the company scale capacity, while Nebius also has a five-year AI infrastructure agreement with Microsoft valued at up to $19.4 billion and commitments from Meta of up to $27 billion.

Today, we typically see several customers competing for every GPU we bring online.

InvestorPlace takes a more selective approach, arguing that the market recently sold AI-linked companies too broadly despite major differences in their fundamentals. Its examples include Micron, Qualcomm, Cerebras, IonQ and Alphabet, though those views are presented as investment opinions rather than guarantees of future returns.

What It Means for You

For U.S. investors, the main takeaway is that buying an AI label is becoming less useful as a strategy. Chipmakers, cloud providers, software companies, storage businesses and networking suppliers can respond very differently to the same spending cycle. That raises the value of examining revenue growth, customer commitments and the specific infrastructure problem each company solves.

Valuation also matters. Another Motley Fool report said Microsoft had fallen more than 20% in 2026 and was trading at 19 times fiscal 2027 earnings, while Amazon was roughly flat for the year despite spending $200 billion on data center expansion. At the opposite extreme, Sandisk had risen hundreds of percent in 2026, showing how dramatically expectations can diverge within the same AI theme.

That does not make any of these stocks automatic buys. The supplied reports contain bullish recommendations and promotional investment opinions, so readers should separate reported company results from forecasts about future share prices.

Coming Up

Several confirmed milestones will shape the next phase of the story. Qualcomm says AI data center silicon revenue will exceed $5 billion by fiscal 2027 and $15 billion by fiscal 2029. Nebius plans to deploy more than 5 gigawatts of capacity by the end of 2030, while Alphabet ended its first quarter with a Google Cloud backlog of $462 billion, most of which management expects to recognize as revenue over the following 24 months.

Those targets will give investors measurable evidence on whether the AI spending boom is broadening beyond chips and into cloud services, storage, networking and other infrastructure layers.

At a Glance

• The Magnificent Seven were flat as a group in 2026 while semiconductor shares and the broader market performed differently.
• Morgan Stanley and Goldman Sachs see potential for renewed interest in large hyperscalers.
• AI infrastructure bottlenecks include advanced packaging, high-speed optical connections and cooling.
• The Nasdaq CTA Artificial Intelligence Index gained more than 60% over the previous year.
• Company results show sharp differences across storage, cloud, software and chip businesses.
• Future revenue targets and capacity expansion plans will test whether AI demand remains broad and durable.