1. Dell Technologies, AI Server Revenue Surges 757% · Annual Forecast Raised to $60 Billion
• Core Source
“AI server revenue surged 757% year-over-year, and the company raised its annual AI-related revenue forecast from approximately $50 billion to $60 billion.”
“We recorded $24.4 billion in AI orders and recognized $16.1 billion in AI server revenue.”
“We are raising our fiscal year 2027 AI server revenue forecast to $60 billion, which shows that the opportunity in the AI market shows no signs of slowing.”
“Demand continues to outpace supply, and memory is the biggest constraint.”
“Driven by growth across Neo Cloud, Sovereigns, and enterprise customers, our AI factory customer count has surpassed 5,000.”
• Expected Impact
Dell Technologies recorded total revenue of $43.8 billion (+88% year-over-year) and AI server revenue of $16.1 billion (+757% year-over-year) in Q1 FY2027 (February–April 2026), overwhelming market consensus (estimated revenue of $34.8 billion). On a backlog basis, $24.4 billion in new orders were added in the quarter on top of the $43 billion AI server backlog at the end of the prior quarter, and 66% of those new orders — $16.1 billion — were converted to revenue within a single quarter. This demonstrates that the speed of conversion from demand to bookings to revenue is exceeding market expectations.
The annual AI server revenue forecast was raised again from $50 billion to $60 billion (+144% year-over-year), and the full-year total revenue guidance was also raised sharply from $138–142 billion to $165–169 billion (vs. consensus of $143.9 billion). Following the earnings release, Dell’s share price surged as much as 40% in after-hours trading.
Notably, traditional server revenue also surged 92% year-over-year, not just AI servers. According to Dell management, Agentic AI — AI that goes beyond generation to independently judge and act — is creating new demand for CPU-based server infrastructure to handle complex branching, memory management, and exception processing beyond GPU compute. This illustrates a structural shift in which AI benefits are expanding from GPU servers to traditional servers and storage.
On the supply chain front, lead times for cutting-edge racks remain as long as one year, and supply constraints persist. Dell’s stated supply shortage priority order is DRAM > NAND > CPU > HDD. This reflects the structural tightness across the memory supply chain, which simultaneously constrains Dell’s further growth potential while also implying meaningful upside if supply conditions ease.
2. Meta Officially Opens Door to Cloud Business Entry in Case of Excess AI Data Center Investment
• Core Source
“Meta CEO Mark Zuckerberg mentioned at the annual shareholders’ meeting that if excess computing capacity arises due to over-investment in data centers, Meta could enter the cloud computing market as a cloud service provider (CSP).”
“Among the four major hyperscalers — Google, Amazon, Microsoft, and Meta — Meta is the only one that is not a CSP, so leaving open the possibility of becoming a CSP provides a justification for setting a higher ceiling on AI CAPEX.”
“Mark Zuckerberg noted that if excess computing capacity arises following Meta’s AI data center expansion, there is a possibility that Meta could enter the cloud computing business. As AI competition intensifies, computing infrastructure is emerging as a core asset that major technology companies want to directly own and control.”
• Expected Impact
Zuckerberg disclosed at the annual shareholders’ meeting that Meta is officially considering entry into the cloud service provider (CSP) business by selling excess computing capacity externally, should its AI data center investment result in over-supply. Among the three major hyperscalers — Google (GCP), Amazon (AWS), and Microsoft (Azure) — all provide external cloud services, while Meta has been the sole hyperscaler without a CSP business.
The reason this statement commands attention goes beyond the mere possibility of entering a new business. CSP entry functions as a buffer against investor concerns about aggressive CAPEX expansion. In other words, if large-scale AI infrastructure investment is redefined not merely as internal AI development spending but as infrastructure assets capable of external monetization, it creates justification for Meta to set a higher ceiling on its AI CAPEX. Meta had previously guided 2026 annual CAPEX at $115–135 billion, a sharp increase from $72.2 billion in 2025.
Combined 2026 CAPEX for the four big tech firms — Google, Amazon, Microsoft, and Meta — is estimated to exceed approximately $600 billion, with roughly 75% concentrated in AI infrastructure. If Meta actually enters the CSP market, it would introduce a fourth major player into the currently oligopolistic global public cloud market. This could have compounding effects: intensified competition with the existing Big Three CSPs, further expansion of demand for GPUs, servers, and networking equipment, and supply diversification options for AI startups and enterprise customers.
3. SanDisk CTO Warning: “The AI Race Has Become a Memory Battle — Shortage to Last Until 2030”
• Core Source
“An unprecedented memory shortage could persist until 2030.”
“Large customers are actively signing LTAs. We have already secured 5 LTAs of up to 5 years each.”
“In all my years as an industry veteran, I have never seen customers this desperate to lock in memory supply volumes.”
“SanDisk announced that it has signed as many as 5 large long-term purchase agreements with global big tech customers, totaling at least $42 billion over a maximum of 5 years.”
• Expected Impact
SanDisk CTO Ilkbahar stated in an interview with Nikkei Asia that the core axis of AI infrastructure competition has shifted from compute (GPU) to memory. As LLM parameter counts scale and AI context windows expand, memory requirements per AI workload are growing exponentially — a trend he characterized not as a normal business cycle but as structural demand expansion.
In practice, SanDisk has signed as many as 5 long-term purchase agreements (LTAs) totaling at least $42 billion over up to 5 years with global big tech companies. This is an unprecedented level of advance contracting never seen before in the memory market, showing that buyers have adopted securing years’ worth of supply visibility as their top strategic priority over short-term procurement. Dell’s COO confirmed on the same day’s earnings call that “memory is the biggest supply constraint,” and TrendForce raised its 2027 global memory market forecast to over $1.28 trillion.
Furthermore, SanDisk announced that it is currently designing dies for HBF (High Bandwidth Flash), a new AI memory specification, with sample supply beginning at end-2026 and a full product including controller slated for 2027. Mizuho raised its SanDisk price target from $1,625 to $1,825, while Barclays raised its target to $2,300.
4. IBM Announces Investment of Over $10 Billion in Quantum Computing Over the Next Five Years
• Core Source
“IBM announced plans to invest more than $10 billion in the quantum computing field over the next five years.”
“The company also reaffirmed its existing target of launching the first large-scale fault-tolerant quantum computer by 2029.”
“IBM (+3.53%) announced plans to invest more than $10 billion in quantum computers over the next five years to build the first large-scale quantum computer capable of performing complex computations stably without errors by 2029.”
“It was reported that IBM is set to receive $1 billion — half of the total — as the Trump administration decided last week to make equity investments of $2 billion in total across 9 quantum computing companies.”
• Expected Impact
IBM announced that it would invest over $10 billion of its own capital in quantum computing over the next five years, reaffirming its target of building a large-scale fault-tolerant quantum computer by 2029. A fault-tolerant quantum computer is a system capable of correcting quantum errors — the chronic weakness of current commercial systems — while stably executing large-scale computations. Once realized, this would enable calculations across industries including cryptanalysis, drug discovery, and financial optimization that are impossible for today’s supercomputers.
The fact that IBM’s announcement coincided with the Trump administration’s decision to make $2 billion in equity investments across 9 quantum computing companies under the CHIPS and Science Act — with IBM set to receive $1 billion, or half the total — amplifies the significance of this development.
The broad rally in quantum computing stocks following IBM’s announcement — IonQ (+7.25%), Rigetti Computing (+9.79%), D-Wave Quantum (+7.31%), Infleqtion (+14.94%) — shows that markets interpreted this investment declaration not as IBM acting alone, but as a signal of ecosystem-wide investment across the U.S. quantum computing industry.
5. Nikon Challenges ArF Lithography Market Share with Price Competitiveness Against ASML
• Core Source
“Nikon CEO Yasuhiro Ohmura stated that Nikon is pursuing new ArF (argon fluoride) lithography equipment orders by offering lower prices than ASML in the semiconductor lithography equipment market.”
“CEO Ohmura explained that Nikon produces a significant portion of its key components in-house, enabling it to achieve price competitiveness.”
“The company is also in discussions with major semiconductor firms in the U.S. and Asia, and noted that some discussions are approaching the purchase order stage.”
“Nikon is one of only two companies globally — alongside ASML — that produces ArF lithography equipment. However, Nikon relies on Intel for 80% of its lithography equipment sales.”
• Expected Impact
Nikon’s new CEO Yasuhiro Ohmura officially declared a low-price strategy against ASML in an interview with Nikkei. The ArF lithography equipment market is dominated by ASML with over 80% market share, and only two companies in the world — ASML and Nikon — are capable of commercially supplying this equipment. Ohmura’s argument is that because Nikon produces a significant portion of its key components in-house, it can achieve a structurally advantageous cost base to undercut ASML on price.
The urgency behind this strategy stems from the vulnerability of Nikon’s current revenue structure. With 80% of Nikon’s ArF equipment sales concentrated in a single customer, Intel, Intel’s struggles in its foundry business have translated directly into Nikon’s deteriorating results. To break free of this dependency, Nikon is actively approaching major semiconductor firms in the U.S. and Asia, with some discussions reportedly approaching the purchase order stage.
ArF lithography equipment is primarily used in mature-node processes (7nm–28nm) rather than cutting-edge EUV, but demand for mature-node semiconductors — power management, analog, and communications chips — is surging in tandem with AI adoption. A single AI server rack contains not only cutting-edge GPUs but numerous mature-node chips as well. If Nikon succeeds in diversifying its customer base using price competitiveness as its weapon, it would introduce a meaningful competitive variable into what has been an ASML near-monopoly in the ArF equipment market, while also providing an alternative supply option for customers seeking to reduce semiconductor equipment supply chain risk.
6. Tesla Breaks Ground on Dedicated Optimus Factory at Gigafactory Texas
• Core Source
“Tesla officially broke ground on a dedicated Optimus factory at the North Campus of Gigafactory Texas, targeting an annual production capacity of up to 10 million units.”
“The factory footprint matches the length of the existing Gigafactory Texas main building, adding over 5.2 million square feet of new industrial space.”
“The Fremont factory will be converted into the initial Optimus production hub following the end of Model S and X production, with mass production scheduled to begin in July–August of this year.”
“The second-generation line at Gigafactory Texas targets mass production by summer 2027, with Musk identifying Optimus as Tesla’s greatest value creation engine.”
“Challenges remain including actuator and sensor supply chain buildout and AI autonomy advancement, but the project is being pursued at a scale of billions of dollars in investment.”
• Expected Impact
Tesla has broken ground on a dedicated Optimus production factory at the North Campus of Gigafactory Texas, targeting an annual capacity of up to 10 million units. The facility spans over 5.2 million square feet — matching the footprint of the existing Gigafactory Texas main building — representing not merely an expansion of production lines but the construction of an independent, robot-dedicated manufacturing hub. Simultaneously, the Fremont, California factory is ending Model S and X production and being converted into the initial Optimus mass production site, targeting a production launch in July–August of this year.
The basis for Musk designating Optimus as “Tesla’s greatest value creation engine” lies in the potential scale of the humanoid robot market. Goldman Sachs projects the global robotics market will grow from $17 billion in 2028 to $37.8 billion by 2035, identifying data accumulation speed and AI autonomy level as the key variables shaping the market. Tesla is structured to directly transplant the real-world AI learning capabilities accumulated through FSD (Full Self-Driving) development into Optimus — a key differentiator versus competitors.
Execution risks remain, however. Actuator and sensor supply chain buildout and AI autonomy advancement are still ongoing challenges, and the 10-million-unit annual target vastly exceeds current global humanoid robot market demand forecasts by tens of times. With billions of dollars in investment being deployed, the actual pace of mass production ramp-up and yield achievement will be the critical monitoring metrics going forward.
Comment [0]