Stocks finish flat on sleepy trading day

Marvell Technology is enjoying another bump in its stock in early trading on Thursday following a series of target hikes from Wall Street analysts and strong readacross from Cisco, which is surging after reporting an earnings beat and a boost to guidance.

Ahead of its Q1 2027 earnings, expected to be released on May 27, Bank of America’s Vivek Arya raised the chipmaker’s price target to $200, from $125, while maintaining a Buy rating on Wednesday. Calling the chipmaker a “top pick,” Arya highlighted the growing potential of AI data center’s total addressable market, or future market size, as well as the role of AI networking — the hardware that powers data transfers between chips, optical components, and servers, which MRVL specializes in, and has been bringing in deals from big clients like Nvidia — in that expansion.

Goldman Sachs analysts took a more cautious stance on a Wednesday note, sticking to its Neutral rating despite bumping its 12-month price target to $125, from $100 previously, as well as hiking its FY27/28 EPS estimates by 5%. The analysts, led by James Schneider, expect “upside to Marvell's Datacenter business driven by higher hyperscaler CapEx, upside in its optical networking business, and a potential new Google partnership,” whilst noting the potential risk in a slowdown in overall AI spending, or the loss market share in custom compute, as reasons for the overall neutral rating.

MRVL is rated as Buy by 86% of the 50 Wall Street analyst recommendations compiled by Bloomberg, with the remaining 7 analysts rating it as Hold. Late Tuesday. Advanced Micro Devices disclosed that it had increased its small stake in Marvell, worth ~$6.5 million at the end of March, in a quarterly filing.

Elsewhere, Cisco jumped on a solid earnings beat and better-than-expected guidance. Both Cisco and Marvell are exposed to the network fabric around AI compute and the data center buildout, but Marvell focuses more on custom chips, while Cisco is exposed to the buildout of switching and routing for AI/GPU cluster networks.

The breathless rally in memory stocks, driven by optimism about the magnitude and longevity of the data center buildout creating a continued supply crunch across the AI supply chain, has put Micron within spitting distance of the $1 trillion market cap mark and given Sandisk a shot at doing something remarkable — topping the S&P 500 two years in a row.

It’s also reshaped where liquidity is deepest on Wall Street, with MU having taken the place of Nvidia as the most-traded stock, while Sandisk, despite being a fraction of the size, isn't far behind.

Technically, Micron’s daily volumes first exceeded Nvidia’s in a single session back on March 19th, but one session isn’t really conclusive evidence to crown Micron king. However, over the last 9 trading days, Micron’s volumes have eclipsed Nvidia’s on 6 of them — and the rolling 5-day average of their turnover now shows clear daylight between the two, with $47 billion changing hands in Micron, compared to just $34 billion in Nvidia, per data from Bloomberg.

Sandisk — which, with a market cap of close to $200 billion, really has little business being in the conversation — is also picking up heat. The stock’s turnover even briefly surpassed that of Tesla, which swapped the crown back-and-forth with Nvidia for much of 2025.

Indeed, when compared to its peers of a similar size (members of the S&P 500 with a market cap less than $250 billion), Sandisk is quite the exception. The company is turning over more than 10% of its market cap on most trading days. The only stock that comes anywhere close to that level is Lumentum, another AI winner.

Doximity plunged in premarket trading after it reported quarterly earnings results that missed Wall Street expectations and gave a disappointing full-year outlook as high AI costs weigh on margins.

For its fiscal Q4 2026, which represents the first three months of this year, the company reported adjusted earnings per share of $0.26, below the $0.28 analysts polled by FactSet were penciling in. Doximity said the profit miss was “driven by AI compute costs.”

And those higher AI costs will follow them into the current fiscal year, which ends next March. For FY 2027, it expects adjusted EBITDA to hit between $323 million and $335 million, lower than the $349.5 million analysts were expecting. Doximity expects FY 2027 revenue to come in between $664 million and $676 million — also below the $682 million that analysts had forecasted.

Doximity, which makes digital tools for healthcare professionals, is building AI products for tasks like medical scribing. Last year, Doximity acquired Pathway Medical, a medical AI startup, for $63 million “and now we're spending against the opportunity it unlocked,” CEO Jeffrey A. Tangney told analysts.

Tangney said the company has “forecasted minimal AI revenue contribution this fiscal year, while allowing for a wider range of AI investments and related expenses, meaning higher R&D, compute and marketing spend, that will weigh on near-term margins.”

“We think that's the right trade,” Tangney said. “This is our AI investment year.”

Lightwave Logic released its Q1 earnings report Wednesday postmarket. The company reported increasing shortfalls as the photonics company continues to scale. Investors reacted by pushing the stock slightly down after-hours.

Here are the numbers: 

• Revenue of $29,000, 27% growing year-over-year.

• Net loss of $6.3 million, widening 34% year-over-year.

The material photonics company, which designs and provides polymers to speed the flow of information from chip to chip, hit a four-year high this week and has risen nearly 400% since January. Daily options volumes on the stock hit a record high ahead of this release.

The stock has been boosted by an explosion of AI data center demand and interest in the growing industry of photonic integrated circuits for data center connectivity.

On their afternoon earnings call, Lightwave Logic CEO Yves LeMaitre reiterated that he believes the company is "positioned to help address some of the most important challenges facing AI infrastructure over the coming decade."