A Blowout Quarter That Lifted All of Tech
The dominant story in semiconductors is a set of earnings results from a memory chipmaker that came in not merely above expectations but above even the highest-end "whisper" estimates circulating ahead of the report. The bar going into the print was extremely high — a fact discussed across multiple shows and segments the day before — yet the company cleared it decisively. Shares rose as much as 18% in early trading on the back of the results.
What makes the move notable is the context. In recent earnings seasons, several technology companies that had rallied into their reports ended up selling off afterward, because investors had set the bar so high that a single weak metric was enough to trigger disappointment. The fear going into this memory report was the same: that even on blowout numbers, the market might fixate on some small data point that "would not resonate" and fail to reflect the genuine strength of the business. The reasonable analytical exercise was to read through the numbers and try to find the metric that wouldn't beat by enough to satisfy the street — and the striking conclusion was that no such metric could really be found. The tone from management clearly worked, and the stock responded accordingly.
The Numbers Behind the Reaction
The raw figures explain the enthusiasm. In the same quarter a year earlier, the company posted $9.3 billion in revenue. This quarter, revenue came in at $41.46 billion, well above the roughly $36.5 billion that had been anticipated — a 345% increase year over year.
Earnings per share were even more dramatic. Adjusted EPS came in at $25.11, against estimates of roughly $20 to $21, and against just $1.91 in the year-ago quarter. That represents a 1,200% increase year over year on adjusted EPS.
A natural question arises from this: with the stock surging, will the price-to-earnings multiple start to creep up to ridiculous levels? The answer is no — and the reason is that the "E" part of the P/E ratio is growing so much and so exponentially that it keeps the multiple in check. Because earnings have been so blowout over the last couple of quarters, the forward-looking P/E has actually stabilized at roughly 10 to 11 times next year's earnings. In other words, the multiple has not run away because the earnings denominator is expanding even faster than the price.
Why This Is a Supply-and-Demand Story, Not a Boom-and-Bust Cycle
The core narrative is one of demand outstripping supply. Historically, the memory chip space has been a classic commodity business defined by sharp boom-and-bust cycles. What is different now is that the company has signed roughly 16 new customers extending their contracts out to around 2030. This transforms what was once a volatile, short-duration commodity market into a longer-duration positive guidance story — converting cyclical unpredictability into multi-year visibility.
Demand has been explosive specifically in everything linked to AI memory — DRAM, NAND, and high-bandwidth memory. Hyperscaler spending has been especially important, continuing to pull forward memory demand across the board. As a result, the company has become a core beneficiary of a broad memory chip shortage, with substantial pricing power. Memory prices continue to rise because demand is outpacing supply, and the business is behaving less like a traditional commodity producer and more like a scarcity-premium AI supplier.
Margins That Beat the Biggest Names in Tech
Gross margins illustrate this scarcity premium vividly. The company guided for about 80% to 81% gross margins; the whisper on the street was around 83% to 84%; and the actual figure came in at 84.9%, nearly 85% — far above historical levels. To put that in perspective, that gross margin is above Meta Platforms, which runs around 82% and relies on high-margin advertising revenue, and well above Nvidia at roughly 75%. By this measure, the memory chipmaker is blowing away the competition.
This was, by all accounts, a flawless quarter. The reaction from analysts confirmed it: a wave of price target hikes followed, some moving up to and above the $2,000 to $2,200 range. The strength is also driving the broader memory space, with competitors such as SanDisk surging as well.
Memory Stocks as the Market's Hidden Engine
Beyond the single company, there is a broader market-structure point. Commentators have frequently noted the lack of breadth in the market — but the memory stock space offers a counterpoint. Names like Western Digital, SanDisk, and Micron have buoyed the overall market and accounted for a huge bulk of this year's gains. These names had all been on a strong run, then pulled back substantially earlier in the week (the weakness was felt particularly on Tuesday), but the renewed surge restored momentum. This kind of participation from the memory complex is precisely what the market needs in order to hold in positive territory. The point connects to a shift in tech leadership: where the hyperscalers and many of the "Magnificent Seven" names have shown weakness, other names such as Oracle and Broadcom have taken over the reins of leadership.
Qualcomm's Reinvention Beyond the Smartphone
The second major story is Qualcomm, whose stock popped significantly — up around 10.5% in early trading — following good news delivered late the prior day at its analyst day. The catalyst is a strategic repositioning away from the handset business that has traditionally been its primary source of revenue. The company is broadening how it intends to make money in the future and raised its guidance in the process.
Central to the story are the words "data center" and "AI-linked data center," which tend to bode well for shares. Qualcomm is making a major push to extend its AI presence while shifting away from its traditional reliance on smartphones. Concrete moves include:
- An acquisition of AI software. The company announced it will acquire AI software firm Modular for roughly $4 billion. The technology allows AI models to run across different chips without rewriting the code. This is strategically aimed directly at Nvidia's single biggest advantage — its software ecosystem and full-stack offering, which is not merely hardware.
- A broader AI data center roadmap, including new CPUs and infrastructure projects.
- Partnerships with major hyperscalers, including a deal with Meta Platforms, which was a key part of the positive news.
Qualcomm has also set aggressive growth targets, aiming for roughly $40 billion in non-handset revenue by 2029 — nearly doubling its prior expectations. The bottom line of the strategy is to reposition the company as a full-stack AI platform spanning chips, software, and cloud infrastructure, rather than being viewed simply as a mobile chipmaker. Part of this broadening footprint also includes the Internet of Things, industrial, and automotive segments, which are driving a meaningful share of gains.
Qualcomm's Performance and the Apple Overhang
Despite the day's surge, Qualcomm has not enjoyed the same outperformance as much of the rest of the semiconductor space, which has been on a winning streak. The stock is up about 15.5% for the year, but it has pulled back roughly 21.5% over the month, having missed out on the broad tech favorability. On a technical basis, it has not slipped below its 50-day or 200-day moving averages and was set to open slightly above those key levels — a hopeful sign for adding to yearly gains, even though June has not looked promising.
A long-running concern hangs over the name: the prospect of Apple taking modem business away from Qualcomm, which it has slowly been doing, with the transition expected to be complete by perhaps 2027. Notably, this potential loss does not appear to be hurting the stock as much as feared. The risk has been discussed for roughly five years, and while it has capped a lot of Qualcomm's gains — especially during periods when other chipmakers were rallying — the company is now showing a clear resurgence.
Espionage in the AI Space: The Anthropic–Alibaba Allegation
The final story shifts to AI competition and national security. US AI company Anthropic has accused Alibaba — the Chinese technology and e-commerce giant — of illicitly extracting knowledge from its Claude models. This development is not surprising; Alibaba is currently suing the United States over having been placed on several blacklists restricting its access to technology. The broader battle between the US and China over technology, and the potential theft of intellectual property, forms the backdrop.
The specifics, according to Anthropic, are striking. Operators linked to Alibaba and its Qwen AI lab allegedly ran a massive, coordinated extraction campaign targeting Claude. The campaign involved roughly 28.8 million interactions with Claude, using about 25,000 fraudulent accounts, over a six-week period spanning from April 22nd to June 5th. Anthropic calls this its largest known distillation attack. This was not normal usage; it was high-volume, automated querying explicitly designed to extract knowledge.
The technique at the heart of the allegation — distillation — is itself a legitimate AI method in which a smaller, cheaper model learns from the outputs of a larger model. What makes this case an alleged attack rather than ordinary practice is threefold: it was done (a) without permission, (b) using fake accounts to bypass safeguards, and (c) at industrial scale. The mechanics are important to understand: rather than stealing code outright, the operators were asking millions of questions, gathering the resulting data, and then using that data to train and compete with rival AI models.
This raises an open-ended challenge that is becoming a recurring theme: how does a leading AI company get ahead of this kind of extraction? It is almost certainly not the first time something like this has happened, but it may be the first instance at this scale and involving companies this large. It is likely to be the first of many such stories, and it reflects exactly the national security concern — agreed upon across both political parties — about being mindful of the technology being exported to China. The episode amounts to large-scale espionage in the AI space, an ongoing front in the US–China technology rivalry that will continue to develop.
