1. NVIDIA F1Q27 Earnings ① Overview: Revenue +85% YoY, 14 Consecutive Quarters of Growth
• Core Source
“Revenue of $81.62 billion (up +85.2% year-over-year) was recorded. This exceeded consensus by $2.65 billion.”
“Non-GAAP EPS of $1.87 (up +238.7% year-over-year) was recorded. This exceeded consensus by $0.10.”
“Data Center $75.25 billion (+92.4% YoY) vs. consensus $73.48 billion — Hyperscale $37.87 billion (+115.2%) / AI Cloud, Industrial, and Enterprise $37.38 billion (+73.7%)”
“Revenue of $91.0 billion vs. consensus $86.95 billion — No contribution from data center compute revenue in China is assumed.”
“Free Cash Flow $48.55 billion (+85.8%) / Operating Cash Flow $50.34 billion (+83.6%)”
• Expected Impact
NVIDIA recorded revenue of $81.6 billion this quarter, extending its streak to 14 consecutive quarters of sequential growth, with the rare pattern of YoY growth actually accelerating compared to the prior quarter. This proves in numbers that AI infrastructure investment has not yet reached its peak.
Data Center revenue of $75.2 billion accounted for approximately 92% of total revenue, and notably, both Hyperscale (+115.2%) and the AI Cloud, Industrial & Enterprise segment grew in a balanced manner. This is a signal that the demand base is broadening away from Hyperscale concentration, demonstrating that a structural diversification of demand is underway.
The Q2 guidance of $91.0 billion exceeded consensus of $86.9 billion by approximately 4.7%, and the fact that China data center revenue was not included at all means additional upside potential remains. If the U.S.-China semiconductor regulatory environment improves, further growth beyond guidance is structurally possible.
Free Cash Flow of $48.6 billion represents an approximately 86% increase from $26.1 billion in the same period last year. The fact that a company with a market cap exceeding $5 trillion generates $48.6 billion in cash on a quarterly basis signals that NVIDIA is transitioning beyond a pure growth stock into a high-quality cash-flow-generating enterprise. This underpins the historic shareholder return policy of an additional $80 billion share buyback authorization and a dividend increase from $0.01 to $0.25 per share.
2. NVIDIA F1Q27 Earnings ② Vera CPU: Opening a $200B TAM
• Core Source
“We have visibility to nearly $20 billion in total CPU revenue this year, positioning us to become the world’s leading CPU supplier.”
“This means NVIDIA is on track to become the world’s largest CPU supplier in its very first year of CPU sales. This is exactly the same outcome as what happened in networking, where NVIDIA grew larger than all competitors combined in just two years.”
“The $20 billion guidance for Vera CPU includes only standalone sales. In AI agents, the GPU handles inference while the CPU handles orchestration.”
“Beyond the $1 trillion guidance, additional revenue streams exist. Standalone Vera CPU sales will become the second largest revenue source after Blackwell/Rubin.”
• Expected Impact
The real surprise of this earnings call was not the GPU — it was the CPU. NVIDIA disclosed that it has already secured $20 billion in revenue visibility for Vera CPU in its very first year entering the CPU market. This is equivalent to AMD’s CPU revenue over the past 12 months, meaning NVIDIA effectively enters the ranks of the world’s largest CPU suppliers from day one.
More importantly, this $20 billion is pure additional upside entirely excluded from the existing Blackwell/Rubin $1 trillion guidance. In other words, a new growth engine has been added on top of growth prospects the market has already priced in.
The logic behind Vera CPU is clear. In an agentic AI environment, AI agents perform CPU-intensive tasks such as orchestration, memory management, and tool calls at massive scale. Jensen Huang explained that “billions of agents will exist in the world, and they will all use tools just like humans use PCs today,” describing a dual demand structure where GPUs handle inference and CPUs handle orchestration. CPU demand is not a replacement for GPU demand — it represents an expansion of the total market.
Vera CPU delivers 1.5x performance per core vs. x86, 2x power efficiency per watt, and 4x rack density, and BofA forecasts the server CPU market will grow from approximately $43 billion in 2026 to $125 billion by 2030 in the agentic AI era. NVIDIA is positioned to capture first-mover advantage in this market, just as it did with GPUs.
3. SpaceX S-1 Filing: Starlink $11.4B vs. AI Losses of $4.9B — The Reality Behind a ‘Contradictory Report Card’
• Core Source
“SpaceX recorded revenue of $18.67 billion last year, growing 33% year-over-year, but recorded a net loss of $4.94 billion due to expanded R&D and AI infrastructure investment. In 2024, the company had recorded a profit of $791 million, reflecting its entry into an aggressive investment phase.”
“In particular, Starlink recorded revenue of $11.4 billion last year, emerging as the core cash cow, while AI segment revenue also grew to approximately $3.2 billion.”
“Anthropic is required to pay SpaceX $1.25 billion per month for computing power rental. This is $3.75 billion on a quarterly basis, far exceeding SpaceX’s Q1 operating loss of $1.94 billion.”
“Given that the Cursor acquisition price is fixed at $60 billion and the capital raised through the IPO is $75 billion, if Cursor is actually acquired, SpaceX’s valuation would inevitably be partially impaired.”
• Expected Impact
SpaceX’s S-1 is exactly what commentators described as “a report card where extremely contradictory indicators coexist.” Starlink generates annual revenue of $11.4 billion as an overwhelming cash cow, but the AI (xAI) segment alone recorded an operating loss of $1.94 billion in Q1 2026. The reason a company that was profitable in 2024 swung to a $4.94 billion net loss in 2025 is a surge in AI infrastructure capex.
However, evaluating the company based solely on surface financials is misleading. The contract under which Anthropic pays SpaceX $1.25 billion per month — $3.75 billion per quarter — in computing rental fees is a new revenue stream not reflected in the S-1 figures, and this alone nearly doubles the Q1 operating loss of $1.94 billion. In other words, an operating profit turnaround from Q2 2026 onward is effectively a foregone conclusion.
Nevertheless, there is a risk investors must watch closely. If $60 billion of the $75 billion raised through the IPO is spent on acquiring Cursor, there is a disconnect between SpaceX’s core narrative and its capital allocation. SpaceX’s enterprise value of $1.75 trillion is premised on the vision of launch cost innovation through Starship and a space infrastructure platform — and if the raised capital flows into a coding AI M&A rather than Starship development, the justification for that valuation is weakened. The figure of $46.9 billion in Q1 revenue relative to enterprise value illustrates how heavily the current premium depends on future growth.
4. Tesla FSD Official Launch in China — Subscription Model Transition and Full Rollout Forecast for Q3
• Core Source
“On May 21, Tesla officially announced that its Supervised Full Self-Driving system, FSD, has been activated in the Chinese market.”
“In the Chinese market as well, the FSD purchasing method is transitioning from ‘one-time purchase’ to ‘subscription (SaaS).’ Expected pricing: The European market level (€99/month, approximately ₩780,000) is expected to be applied, with discounts for existing EAP (Enhanced Autopilot) holders.”
“As Elon Musk mentioned in the Q1 earnings call, the target is to obtain final approval from Chinese authorities and achieve full deployment within Q3 2026.”
“Following Trump’s visit to China, in which Musk CEO accompanied, Tesla posted job listings urgently seeking specialized personnel to test and analyze autonomous driving operations in more than 9 major cities in China.”
• Expected Impact
The significance of this China FSD launch is not merely a feature expansion. The key is that a structural change in the business model — transitioning from one-time purchase to subscription (SaaS) — is being applied simultaneously. A monthly subscription model generates Recurring Revenue separated from hardware sales, signifying a transition from an automotive company to a software platform company.
China is a critically important market for Tesla. When FSD is fully commercialized in the world’s largest EV market, the large-scale Chinese driving data Tesla possesses will be used to train local navigation algorithms, forming a virtuous cycle in which autonomous driving performance improves exponentially. The fact that Tesla is urgently recruiting specialized test personnel across more than 9 cities already signals that localization work for full Q3 deployment is underway in earnest.
However, the fact that L3 and above autonomous driving services remain uncertain in outlook is a constraint. The currently launched ‘FSD Supervised’ is at the L2 level requiring driver attention, which invites comparison with competitors already operating robotaxi services in China. The subscription conversion rate and actual monetization speed following full Q3 approval will be the key variables to watch.
5. U.S. Invests $2B in 9 Quantum Computing Companies via CHIPS Act — World’s First Quantum Chip Foundry to Be Established with IBM
• Core Source
“The U.S. government plans to provide a total of $2 billion in support to 9 quantum computing companies and conduct minority equity investments in parallel. IBM is included as a recipient of $1 billion, and GlobalFoundries of $375 million. D-Wave, Rigetti Computing, Inflection, Atom Computing, PsiQuantum, Quantinuum, and others are each expected to receive $1 billion in support. Dirac is included as a recipient of $38 million. Funds are to be disbursed through the CHIPS and Science Act, with the support structure involving grants and minority equity investment by the U.S. government. However, the transactions have not yet been finalized.”
“IBM and the U.S. Department of Commerce announced plans to establish America’s first pure-play quantum chip foundry through a new company called Anderon. This project is to be pursued with a proposed $1 billion CHIPS grant alongside $1 billion in investment from IBM comprising cash, intellectual property, assets, and personnel. The foundry will support quantum wafer production for multiple quantum hardware companies, with the goal of strengthening the U.S. quantum computing ecosystem.”
• Expected Impact
This measure has two core elements. First, the U.S. government is adopting a ‘minority equity investment’ structure in parallel with grants. This signifies that the government has officially acknowledged the monetization potential of the quantum computing industry, and demonstrates that the CHIPS Act is expanding beyond semiconductors to encompass next-generation computing infrastructure broadly.
Second is ‘Anderon,’ the world’s first pure-play quantum chip foundry being established as a joint venture between IBM and the Department of Commerce. Funded at a total of $20 billion (government $10 billion + IBM $10 billion), this foundry will provide quantum wafer production services not to a single company but to multiple quantum hardware vendors. This can be viewed as the beginning of a structural transition in which the semiconductor foundry ecosystem is replicated in the quantum computing domain.
From an investor perspective, the noteworthy point is that immediately following this announcement, IonQ (+8.32%), D-Wave Quantum (+6.10%), Rigetti Computing, and other related stocks surged in unison. The government’s adoption of an equity investment structure serves as a public endorsement of the technological credibility of these companies, catalyzing inflows of private capital. Note, however, that the transactions have not yet been finalized.
Quantum computing has the potential to process in minutes what would take conventional supercomputers thousands of years — in AI training and inference, drug discovery, cryptography, and more. Just as the current AI boom created a GPU infrastructure investment cycle, the commercialization of quantum computing could become the seed of the next infrastructure investment cycle, making the long-term significance of this government-led ecosystem development substantial.
6. ASML CEO: “Supply Chain Bottlenecks Observed at Multiple Stages, No Resolution in Sight”
• Core Source
“ASML’s CEO Christophe Fouquet stated that the global semiconductor supply shortage is intensifying and presented the outlook that this problem will not be resolved for the foreseeable future.”
“The global semiconductor market is growing rapidly and is expected to reach $1.5 trillion by 2030. However, bottlenecks are being observed at multiple stages of the semiconductor supply chain.”
“Demand for advanced edge wafers in 2027 is expected to increase by 25–30%, while ASML’s production capacity provides conditions under which wafer supply can be expanded by more than 50%.”
“ASML’s memory-related revenue is expected to account for 30–35% of its 2026 revenue, exceeding the semiconductor equipment sector peer average of 25–30%. It can be expected that ASML will benefit more from the memory supply shortage.”
• Expected Impact
ASML’s CEO’s remarks are a structural diagnosis of the entire semiconductor supply chain. The key point is that bottlenecks are occurring simultaneously across ‘multiple stages’ of the supply chain, not merely at a specific process or company. AI demand is rising so steeply that equipment lead times, wafer processing capacity, and advanced packaging — each link in the chain — cannot keep pace.
ASML’s direct beneficiary logic is clear. Demand for advanced edge wafers is expected to grow 25–30% in 2027, while ASML itself stated that its equipment can expand wafer supply by more than 50%. This means ASML is not a victim of the current supply chain bottleneck but rather the key beneficiary holding the solution to resolving it. UBS raising its target price from €1,600 to €1,900 reflects this logic.
Two additional points deserve attention. First, the forecast that the first logic chip production based on High-NA EUV will begin within a few months signals the point at which demand for ASML’s next-generation equipment begins to materialize in earnest. Second, ASML is expanding its business scope beyond lithography equipment into advanced packaging equipment. As the scaling of AI chips makes the packaging process a central battleground for performance competition, ASML’s TAM is structurally expanding.
Comment [0]