This story originally appeared in The Technology Letter and is republished here with permission.

When an analyst leaves a bank to join another bank, it’s a chance to make a big pronouncement about how they view the overall industry they cover.

Such was the case over the weekend as former Barclays analyst Blayne Curtis took over semiconductor coverage at Jefferies & Co. from Mark Lipacis (who has gone over to Evercore ISI).

Curtis starts his coverage of Nvidia and several other names with a very positive view, arguing that it is “Still early in the cyclical recovery and overall group can still work.” We actually heard the same view last month from Curtis’s colleague who covers European semiconductors, Janardan Menon.

Nvidia is Curtis’s favorite. He focuses quite a bit on the company’s new “Blackwell” GPUs introduced in March. He’s particularly interested in what’s known as the “GB200 NVL,” which is a rack-mounted system in which seventy-two Blackwell GPU chips are joined to thirty-six of Nvidia’s “Grace” CPUs. That product is going to give Nidia far more power over the architecture of artificial intelligence in the data center, Curtis believes.

“This would allow NVDA to control even more content decisions, particularly in the early ramp,” writes Curtis.

Along with Nvidia, Astera Labs, the recently public vendor of memory-bandwidth technology, which last week reported a good quarter, is going to be a big beneficiary of the GB200NVL, writes Curtis. Although Astera doesn’t talk about specific customers, he expects more of Astera’s “compute express link,” or, CXL technology, to be designed into the GB200. The GB200NVL is “Positive for ALAB given potential new content in the GB200 NVL,” he writes.

Curtis likes AI as an investment theme overall, writing, “As long as cloud providers continue to foot the bill, the AI group should continue working.” It’s “too early to sift out winners,” he declares, but Advanced Micro Devices also has a shot here. “There is an opportunity for AMD to be a second source GPU option and ultimately that is all that will matter for our Buy rating to be correct.”

Another big theme is buying undervalued analog chip names. These stocks are going to play out well in 2026 and 2027, he believes. “Better early than late with risk/reward favorable with analog stocks flat YTD and fundamentals troughed,” writes Curtis. Among analog chip stocks, Curtis likes large-cap names ON Semiconductor, NXP, and Analog Devices. Among smaller stocks he likes Macom Technology Solutions, Lattice Semiconductor, Allegro MicroSystems, and Impinj.

In the chip-equipment market, “growth in the Semi market (driven by AI) should translate to higher WFE,” writes Curtis, referring to “wafer fab equipment,” the main designation of chip equipment. He favors KLA and Lam Research, but also likes smaller names Onto Innovation and Camtek, who, as I mentioned in the podcast, are both makers of technology for so-called 3-D chip packaging, which is all the rage for large AI chips for the data center. “The advanced packaging mega trend is still early innings,” writes Curtis.

I think most of what Curtis discusses makes a lot of sense. As I mentioned Monday morning, hardware is winning in the AI arms race versus software, as we are in a time of basic infrastructure building in advance of applications. I think Curtis’s stock picks are pretty good, though, as I’ve mentioned before, Astera is frighteningly expensive (not that the current market minds high valuations much.)

The pick of KLA and Lam makes sense, given AI’s need for more and more memory chips; both companies are dominant in sales of equipment to make memory circuitry.

D-Wave Quantum breaks a bad streak

The very young field of quantum computing has so little revenue that it hardly seems to matter what a company reports in that regard. Thus, as D-Wave Quantum Monday morning reported its ninth consecutive quarter of missing analysts’ revenue expectation by a wide margin — $2.5 million in sales versus $3 million expected — the bulls on the stock looked past the matter, and the shares closed the day up over five percent. It was the first time in seven quarters that the stock has not massively sold off following results. The bright spot was the value disclosed for the company’s new contracts, known as “bookings,” which was $4.5 million. Needham & Co.’s Quinn Bolton, who has a Buy rating on the stock, notes that the bookings included “two substantial multi-year orders.” The deals included “new customer use cases,” he notes, such as “solar panel usage in buildings, body shop scheduling for commercial vehicle production, and optimizing schedules for crude oil tanker unloading at refineries to name a few.”

As noted in my interview last month with D-Wave’s CEO, Alan Baratz, the company considers bookings “a much better indicator of the health of the business than revenue”  because of the fact that D-Wave’s professional services dominate at the moment, while actual cloud usage of the D-Wave machines is slower to “ramp up.” It’s really early in this industry, and in D-Wave’s life as a business, to take any quarter’s contracts too seriously. It will take many more quarters to prove that the professional services can turn into a meaningful stream of revenue.

Pick the best platforms in cybersecurity

You may recall that back in February, the CEO of cyber-security firm Palo Alto Networks, Nikesh Arora, remarked that his customers were expressing “fatigue” about spending money on the technology. That sentiment was on display last week at the big annual trade show for security, the RSA Conference in San Francisco, according to Canaccord Genuity analyst Mike Walkley, who was in attendance, chatting with dozens of companies. Walkley came away “incrementally cautious on near-term trends given the tough upsell environment,” and heard that there is “greater deal scrutiny,” which is likely to continue through the first half of the year. At customers, he observes, “teams are encouraged to do more with less following years of significant security investments leading to possible spending fatigue.” Customers still view security software as “must have,” he relates, but it’s getting harder and harder to sell more stuff to each customer.

Walkley points to the decline in “dollar-based net retention,” a non-GAAP number representing the increase in spending, for vendors Tenable Holdings, Qualys, and Rapid7. There is an “increasing customer preference for vendor consolidation,” he notes, the same thing that Arora was talking about. To that end, Walkley advises focusing on those companies that represent “best platforms.” He writes, “Coming out of the conference, we recommend CyberArk, CrowdStrike, and Zscaler as core long-term portfolio holdings given leadership positions in targeted markets and beneficiaries of long-term platform consolidation.” The first two stocks have done poorly this year, CyberArk down two percent, Zscaler down twenty-one percent. CrowdStrike is a stand-out, up twenty-five percent.

Walkley’s points are echoed Monday by Jefferies & Co.’s Joseph Gallo, who spoke with a cyber-security consultant recently. The consultant relates that they have “seen a lot of cyber teams ‘fatigued’ by a combination of a talent shortage in security and too many vendors/tools, driving demand for a single pane of glass.” Makes me wonder when the great consolidation will happen in this industry, which I have been expecting for some time. Gallo, like Walkley, notes that “CRWD & CYBR remain unobstructed winners in our expert’s view.”

Sorry state of iPhone sales a continued focus

Despite an upbeat report from Apple this month that boosted the stock, the weakness of iPhone sales continues to be a key concern of analysts. KeyBanc analyst John Vinh on Monday offers up the conclusions of a survey of carrier sales trends. Observations include that “iPhone 15 sell-through in April was tracking in line with to slightly below store expectations.” The more-expensive models, the iPhone 15 “Pro” and “Pro Max” were “relatively resilient,” he relates, “but the base models were still soft.” Sales in the U.S. in the month of April were down twenty-one percent, month to month, relates Vinh, and down fourteen percent, year over year, which is a reversal of the five-year April sales pattern average for a three-percent gain. Moreover, Apple’s comments that this quarter it expects total revenue to rise “low-single digits” means that iPhone sales will be down this quarter by “mid-single digits,” estimates Walkley. He has a Sector Weight rating on Apple stock, which is down three percent this year.

Apple’s struggle with the iPhone 15 is an old story, of course, and by now, no one cares as much as they care about what kind of AI is going into an iPhone 16. Perhaps the more interesting question is whether AI offerings, such as the rumored partnership with OpenAI, are going to lead to a services component for iPhone that will finally diminish the importance of hardware unit sales growth for the company. I’m not holding my breath, but we’ll know more next month with Apple’s developer conference in San Francisco.

This story originally appeared in The Technology Letter and is republished here with permission.