In the AI race where every company is scrambling to build models, there's one thing everyone has to buy before they can even start playing — the accelerator. And it's sold the way a store sells to whoever walks in, not as a secret chip the tech giants build only for themselves. This lesson walks through why training AI takes thousands or tens of thousands of GPUs running in parallel, why a single company can take nearly 90% of this market with a “moat” made of software, and how far the challengers — AMD and Intel — can break in.
S&P 500 earnings are expected to increase by +24% from the same period last year in the third quarter, the 8th straight quarter of double-digit earnings growth for the index, according to Zacks Investment Research. Earnings are expected to be above the year-earlier level for 14 of the 16 Zacks sectors, with 5 sectors expected to enjoy double-digit growth: Aerospace up +159.3%, Energy up +111.9%, Tech up +41.9%, Basic Materials up +31.2%, and Transportation up +15.1%. The Conglomerates sector is the only one expected to have lower earnings in Q3 relative to the same period last year, down 35.4%, while Consumer Staples earnings are expected to be flat. Excluding the Energy sector, Q3 earnings growth for the S&P 500 drops to +20% from +24%, and excluding the Tech sector, growth for the rest of the index drops to +14.4%. Nvidia's Q3 earnings are expected to increase +90% year-over-year on +91.2% higher revenues, while Micron's year-over-year earnings and revenue growth rates are expected to be +938% and +348.6%, respectively, and Tech sector earnings growth gets cut by slightly more than half once contributions from Nvidia and Micron are excluded. The Q3 earnings season will get the spotlight when the big banks report on October 13th, but the reporting cycle actually got underway with the September 10th quarterly releases from Oracle and Adobe, followed by homebuilder Lennar as the third S&P 500 member to report such Q3 results, with an additional six index members on deck this week including Costco, AutoZone and Darden. Total Q3 earnings for the three S&P 500 members that have reported results already are up +22.6% from the same period last year on +14.9% higher revenues, with 33.3% beating EPS estimates and 66.7% besting revenue estimates.
Citrini Research Finds Humanoid Robots Advancing Without a Single ChatGPT Moment
Humanoid robots are making meaningful progress in autonomy, dexterity and repetitive real-world tasks, but widespread commercial deployment remains a work in progress, according to a report published Wednesday by Citrini Research. After a week meeting robotics labs, engineers and investors across the San Francisco Bay Area, the firm found most humanoid robots remain in development, training and tele-operated demonstrations, with real-world deployments closer to pilots than clear return-on-investment opportunities. The report examined whether robotics is approaching its own ChatGPT moment and highlighted advances from GeneralistAI and Skild AI, along with Figure AI's demonstration involving 250,000 packages, suggesting the field may advance through a mosaic of signals rather than one defining breakthrough. Basic humanoid locomotion is effectively a solved problem while complex dexterity remains more difficult, and simple repetitive tasks are already trainable. Investment in the broader robotics infrastructure is keeping pace: Tesla Inc has been preparing production lines for Optimus, XPeng Inc has launched an automated production line for its IRON humanoid robot, and Figure AI is securing access to up to 100,000 Vera Rubin GPUs from NVIDIA Corp to train the models powering its humanoid robots. The report concluded robotic autonomy is developing along a curve rather than arriving through one decisive breakthrough, with broad generalizability remaining the longer-term goal.
Nvidia CEO Huang Says Chip Sales Will Double Next Year
Nvidia CEO Jensen Huang said the company expects to sell roughly twice as many chips next year as it does this year, calling it one of his strongest signals yet that AI demand remains far from exhausted. Speaking to reporters Thursday at a summit in Scotland with King Charles III, Huang said the reason is AI's contribution to different industries and economies, and that people in almost every country Nvidia operates in want to invest in AI. The forecast adds to Nvidia's increasingly aggressive outlook: the company recently said it expects roughly 70% growth in the fiscal year ending January 2028, which would put annual revenue around $673 billion. Nvidia does not disclose total chip unit shipments, making Huang's volume forecast difficult to translate directly into revenue, though its most important products include data-center GPUs, particularly Blackwell and next-generation Rubin systems; Huang said last fall that Nvidia had shipped about 6 million Blackwell GPUs over four quarters. The comments come as investors debate whether hyperscaler spending and model-training demand can keep growing at the breakneck pace seen during the first phase of the AI buildout, with the key question being not simply whether Nvidia can double chip volumes but what that mix looks like, since higher shipments of premium Blackwell and Rubin systems could support enormous revenue growth while greater unit volume at lower average selling prices would produce a different margin outcome.
Xeal Launches Laitent, World's First Edge Inference Compute Network Using Idle EV Charging Capacity
Xeal launched Laitent, which it calls the world's first edge inference compute network using idle EV charging capacity, tapping more than 200MW of permitted, installed electrical infrastructure across 1,600+ properties. Xeal, a member of NVIDIA Inception, plans to deploy over 100,000 NVIDIA GPUs alongside EV charging infrastructure, and has secured partnerships with Rafay Systems for AI infrastructure orchestration, Spectrum Business for dedicated enterprise-grade fiber, dozens of real estate and property managers, and a Tier 1 inference provider for up to 5MW of compute. The first Laitent Pod will be brought online with partner JVM Realty by the end of 2026. Each Laitent Pod is about the size of one parking space, contains up to 48 NVIDIA Hopper or Blackwell Ultra GPUs, requires no water hookup, and runs quiet at less than 65 decibels, offering sub-20ms latency in metro areas. Xeal said EV charging sites typically operate at less than 10% of permitted capacity, and it taps the remaining 90% for compute, with property owners able to add as much as $1m in property value for little-to-no upfront investment. Looking beyond the initial 200MW of installed charging capacity, Xeal plans to unlock over 1GW of existing headroom across real estate and EV charging deployments.
Nvidia CEO Jensen Huang Signals Chip Sales to Double Next Year
Nvidia CEO Jensen Huang said Thursday he expects the company to sell twice as many chips in the next year as it did this year, a signal to investors that demand has not yet peaked. Huang attributed the outlook to AI investment spreading across industries, economies and countries. Management said consumer expectations point to another doubling of growth next year, and Nvidia projected around 70% revenue growth for the fiscal year ending January 2028, bringing revenue to almost $673 billion. Management classified the outlook as supply tight, implying Nvidia's capacity to ship chips could remain one of the largest impediments to growth. Last year, Huang said the company shipped around 6 million Blackwell GPUs in four quarters, and Rubin is now coming up as the next big part of the product cycle.
Goldman Sachs has pushed back on fears that U.S. corporate profits are in an "earnings bubble," even as it acknowledged some companies are "over-earning." Analyst Ben Snider noted that S&P 500 earnings per share grew 51% in the second quarter and 26% over the past four quarters, lifting profits well above their long-term trend. He added that while the index's forward price-to-earnings multiple of 19 is in line with its 10-year average, its multiple on trend earnings has been exceeded in recent decades only at the peak of the dot-com bubble. Still, Snider said, "our base case is for S&P 500 earnings growth to decelerate, not collapse, in coming years," forecasting EPS growth of 11% in both 2027 and 2028, to $415 and $460, respectively. The biggest long-term question is AI, he said, noting the investment boom has driven nearly half of this year's earnings growth, a tailwind that should fade in 2028 as capital spending slows and depreciation mounts, turning into a marginal drag. Snider also flagged risks from a potential fall in semiconductor prices, where an adverse scenario could cut S&P 500 earnings by about 10%, and from mega-cap technology firms' equity investment gains, which lifted second-quarter earnings by 12% but should diminish in 2027. He concluded that Goldman's 12-month S&P 500 return forecast of +14% to 8,700 reflects the view that earnings growth, rather than expanding valuations, will remain the primary driver of the bull market.
US Approves $24.3B F-35 Sale to Saudi Arabia; China Tariff Decision May Slip
The State Department approved a potential $24.3B sale of 48 Lockheed Martin F-35 Lightning II aircraft to Saudi Arabia, along with 49 Pratt & Whitney F135-PW-100 engines and related equipment, a deal that still requires congressional approval and would strengthen Saudi Arabia's air defense and interoperability with U.S. and allied forces. Separately, Bloomberg reported that the U.S. could postpone announcing new tariffs on China and other countries until after President Donald Trump meets Chinese President Xi Jinping next week; a previously planned trade report was expected to recommend a 7.5% tariff on Chinese goods, while the final rate could potentially exceed the 20% level agreed during earlier talks. Nvidia CEO Jensen Huang said the company expects to sell roughly twice as many chips next year as this year as demand for AI infrastructure expands. China's ChangXin Memory Technologies is preparing to enter the NAND flash-memory market, with Reuters reporting that CXMT plans an R&D production line at its new Beijing plant and has established a research institute in the city, broadening its focus beyond DRAM. OpenAI has hired Brian McCarthy from SpaceX as vice president of worldwide sales, Fortune reported; McCarthy joined SpaceX through its acquisition of Cursor, where he led global revenue and field operations, and previously held senior sales roles at Rubrik, ThoughtSpot, AppDynamics and Qlik.
Nvidia CEO Huang Says AI Spending Flywheel Is 'Really, Really Flying'
Nvidia CEO Jensen Huang said demand for artificial intelligence computing is accelerating, describing the resulting investment cycle as a flywheel that is "really, really flying." Speaking with CNBC's Jim Cramer at Dreamforce in San Francisco, Huang estimated that a 1-gigawatt Nvidia AI factory costs about $50 billion to $60 billion to build but can generate roughly $50 billion in annual rental revenue, implying a return on invested capital of about one year. He said token generation has increased 25-fold in less than a year, with open models now accounting for nearly 70% of generated tokens, up from about 30%. Huang also said Nvidia Grace Blackwell systems now rent for about $16 per GPU hour, versus roughly $5 under some contracts signed a year ago, and he expects AI testing to become a third major category of infrastructure alongside training and inference, requiring additional data centers.
AI Risk Debate Splits Tech Leaders as Traders Weigh Capex Slowdown
A fresh global debate over artificial intelligence risks intensified this week as tech executives clashed over whether frontier labs should slow development of the technology. Leaders of Anthropic, OpenAI, SpaceX and Google DeepMind pushed for self-regulation checks on growth that give AI models more control, while Microsoft, Amazon and Meta said they are building tools to check, control and regulate AI risks, and US President Donald Trump dismissed the warnings as a "hoax." The call for caution triggered a rotation out of tech stocks at the start of the week as investors questioned whether the pace of the booming AI buildout needs to be repriced. Market participants said any sign of a slowdown in AI capex would be a major moment, with semiconductors likely hit first, though several voiced confidence that the spending cycle itself remains intact.
Xi-Trump Summit to Put AI Rivalry, Tariffs and Yuan in Focus
Artificial intelligence, a lingering trade war and the yuan will likely dominate the agenda of the upcoming US-China summit between US President Donald Trump and Chinese counterpart Xi Jinping, with analysts focused on whether the two nations can reach concessions or put guardrails around key areas of competition. A Goldman Sachs Group Inc. survey shows 46% of offshore and 38% of onshore investors expect Chinese stocks to rise after the talks, while only a small minority foresees losses, though overseas exchange-traded fund flows and options positioning suggest foreign investors remain wary ahead of the summit, according to Tony Lee, JPMorgan Chase & Co.'s equity-derivatives strategist. AI is widely expected to be a central topic, with access to advanced US chips, safety standards and disputes over the pace of the industry's development among key points of contention, while the fate of a soon-to-expire one-year tariff truce, Beijing's export curbs on rare earths and the value of China's currency may also be in the spotlight. Expectations are high that the summit may result in an extension of the bilateral trade truce that will expire in November, after the two nations have started talks over slashing tariffs on certain goods, including American energy and agricultural shipments, as well as lower duties on Chinese inputs for manufacturers. The yuan may also be discussed, especially after US Treasury Secretary Scott Bessent said in August that many people considered the Chinese currency undervalued; the yuan has strengthened around 4% against the dollar this year, making itself Asia's top performer.
Nvidia Launches CUDA-Q Logical for Fault-Tolerant Quantum Computing
Nvidia launched CUDA-Q Logical, an open-source orchestration layer built into its CUDA-Q platform, on September 14. The tool is designed to help researchers develop applications for fault-tolerant quantum computers using logical qubits, and it is available on GitHub. Fermi National Accelerator Laboratory, IQM Quantum Computers, Sandia National Laboratories, Infleqtion, QCDesign, and Quantum Motion are already using the technology. The launch supports Nvidia's broader strategy of making its GPUs the key classical computing layer for hybrid quantum-classical systems, though Jensen Huang said in January 2025 that useful quantum computers could still be 15 to 30 years away, a view tied to the unresolved challenge of scaling quantum hardware. Hedge fund ownership of Nvidia rose from 275 funds at the end of Q1 2026 to 285 funds at the end of Q2 2026, while short interest stood at 1.29% of float as of August 31, 2026.
Huawei Pulls Forward Ascend 960DT Launch to Early 2027 as Nvidia Gains 2.5%
Huawei has pulled forward its next generation of AI chips, with the Ascend 960DT now slated for an early 2027 launch, roughly three quarters earlier than previously expected, followed by the 960PR in the third quarter, according to Reuters. Nvidia shares gained approximately 2.5% to $219.24 on the news. Huawei's Peerium architecture is designed to eventually link as many as one million processors, while a nearer-term supernode would connect 4,096 chips, directly targeting the system-level advantage Nvidia has built around GPUs, networking and tightly integrated AI infrastructure. Huawei says production still cannot satisfy demand inside China, making manufacturing capacity a potential brake on both domestic adoption and any meaningful overseas expansion. Nvidia's data-center revenue reached $89 billion, accounting for roughly 92.5% of the company's latest $96.2 billion quarter, and it carries a GF Score of 95 out of 100.
CoreWeave Deploys NVIDIA Vera Rubin NVL72 Clusters on Its Cloud
CoreWeave deployed multi-rack NVIDIA Vera Rubin NVL72 clusters on CoreWeave Cloud on Wednesday, moving the architecture from validation toward broader availability. CoreWeave had already become the first cloud provider to bring up and validate NVIDIA's Vera Rubin NVL72, and in August it won a multi-year agreement with Hudson River Trading to build its next-generation AI research and model-development platform using Vera Rubin NVL72 and Spectrum-X Ethernet. NVIDIA says Rubin is in full production and is designed for training and inference. CoreWeave said AI products and services beyond GPUs, including storage, CPU, networking and software, exceeded $400 million of annual recurring revenue in the second quarter, while booked annual recurring revenue for managed inference rose from $1 million to more than $100 million within months of launch. Active power exceeded 1.5 gigawatts in the second quarter and contracted power reached about 4.2 gigawatts by Aug. 11, with CoreWeave targeting more than 1.85 gigawatts of active power by year-end 2026 and at least 8 gigawatts by 2030, against expected 2026 capital expenditures of $35-$39 billion and second-quarter net interest expense of $640 million.
Meta Could Save $8.5 Billion in 2027 on Custom MTIA Chips, BofA Estimates
Bank of America estimates Meta Platforms could save roughly $8.5 billion in 2027 by running AI workloads on its own custom silicon instead of buying third-party chips, an outside analyst estimate rather than company guidance. The figure rests on a specific roadmap: Meta plans to deploy its third-generation MTIA 450 chip, code-named Arke, in the first half of 2027, followed by the higher-performance MTIA 500, or Astrid, later that year, both co-developed with Broadcom and aimed at AI inference workloads. BofA models Meta deploying 5 to 6 gigawatts of owned capacity in 2027 at a total cost of roughly $200 billion, assumes chips make up 60% of that spend, and pegs Meta's custom silicon as about 40% cheaper than third-party equivalents. Broadcom CEO Hock Tan said custom chips optimized for a customer's own workloads outperform any GPU and can do so at half the cost, and confirmed Broadcom will deliver three generations of MTIA accelerators to Meta between now and the end of 2027. Meta's FY2026 capex guidance sits at $130 billion to $145 billion, narrowed from $125 billion to $145 billion, with total expense guidance raised to $165 billion to $169 billion, while Q2 2026 revenue reached $60.80 billion, up 27.96% year over year, on advertising revenue of $59.36 billion.
Damodaran Says Nvidia Priced as 'Greatest Company Ever'
NYU Stern finance professor Aswath Damodaran said Nvidia Corp. is an awesome company but is being priced as the greatest company ever, arguing the stock demands an extraordinary payoff from the artificial intelligence boom. Damodaran, who bought Nvidia in 2018 at a split-adjusted $1.80 a share and later fully exited the position, now estimates AI product and service revenue of roughly $250 billion a year against more than $2 trillion already invested in AI infrastructure, and says those revenues may eventually need to reach $8 trillion to $10 trillion annually to justify the buildout, roughly 32 to 40 times today's level. He argues AI only approaches that scale if it moves beyond helping workers and starts replacing them, noting companies worldwide pay roughly $26 trillion a year in wages and other employee compensation; if AI remains primarily a productivity tool, he puts the potential market closer to $2 trillion. Nvidia's latest quarterly results showed revenue of $96.2 billion, up 106% year over year, with Data Center revenue up 117% to $89 billion, and Polymarket gives Nvidia a 71% chance of finishing 2026 as the world's largest company versus 20% for Apple Inc., with more than $7.3 million traded. Nvidia's roughly $5.17 trillion market value leads Apple's $4.85 trillion by about $320 billion, so a relative move of around 6% would flip the ranking.
Huawei Accelerates AI Chip Roadmap, Sets 2027 Launch for Two New Chips
Huawei is accelerating its push into artificial-intelligence computing, with two new AI chips now scheduled for 2027 as the Chinese technology giant tries to narrow Nvidia's advantage in one of the industry's most strategically important markets. Rotating chairman David Wang said at Huawei Connect in Shanghai that the company plans to launch its 960DT chip in the first quarter of 2027, followed by the Ascend 960PR in the third quarter, an acceleration from its previously disclosed roadmap that had placed the Ascend 960 launch in the fourth quarter of 2027. Huawei is also building UnifiedBus, an interconnect technology designed to allow large numbers of AI chips to operate together as one larger computing system, and has developed 11 chips around UnifiedBus for its supernode and supercluster systems. The company has already shipped more than 1,000 supernode systems to more than 370 customers, systems that combine multiple accelerators to handle workloads requiring substantially more computing power than a single chip can provide. The bigger long-term risk for Nvidia is geographic: if Huawei continues improving its chips and cluster architecture, it could strengthen China's ability to build AI infrastructure without relying on Nvidia hardware and create a more credible parallel AI-computing ecosystem inside China.
CoreWeave Deploys Multi-Rack NVIDIA Vera Rubin NVL72 Clusters
CoreWeave is expanding its capabilities for next-generation AI workloads with the deployment of multi-rack NVIDIA Vera Rubin NVL72 clusters on CoreWeave Cloud, alongside two new capabilities for its AI Object Storage platform: cross-region write acceleration and a lower-cost Archive tier. The multi-rack Vera Rubin NVL72 deployment allows hundreds of NVIDIA Rubin GPUs to operate as a single scale-out cluster, with each NVL72 rack combining 72 Rubin GPUs with 36 Vera CPUs, NVIDIA NVLink 6, ConnectX-9 SuperNICs and BlueField-4 DPUs, connected across racks using NVIDIA Spectrum-X Ethernet networking and automated through CoreWeave's Mission Control platform. CoreWeave says its Local Object Transport Accelerator caches data locally and can reduce latency by up to 8x compared with traditional storage-cluster reads while delivering up to 7 GB/s of throughput per GPU, while the new cross-region write capability lets customers write data locally as CoreWeave replicates it to another region in the background. The company faces intensifying competition in the AI cloud market from Microsoft and Nebius Group N.V., which plans to invest £1.7 billion in U.K. AI infrastructure and expects connected power capacity to reach roughly 800 MW to 1 GW by 2026 end. CoreWeave shares have gained 24.4% year to date against the Internet Software industry's fall of 0.6%, and the Zacks Consensus Estimate for CRWV's earnings for the current year has been revised downward over the past 60 days.
Nvidia's "Bank of AI" Role Expands as Huang Targets AI's Capital Bottleneck
Nvidia is increasingly acting as what Chamath Palihapitiya called the "bank of AI," working across suppliers, infrastructure providers and financing partners to clear bottlenecks before they constrain AI deployment, Jensen Huang said at the All-In Summit. Huang described AI as a "new industrial revolution" that requires manufacturing, electricity, internet infrastructure and physical data-center capacity to scale together, not just GPUs. He said roughly $400 billion in venture financing flowed into AI companies over six months, creating jobs and massive demand for compute, which in turn drives demand for data centers. Nvidia has increasingly worked with financial institutions and other ecosystem players to make AI compute an investable, financeable asset, a strategy that turns the company into an ecosystem enabler: help customers secure capital, power and infrastructure, and Nvidia keeps supplying the computing engine underneath them. For investors, that could make Nvidia's AI opportunity considerably broader than its semiconductor market share, since the biggest constraint on AI spending may eventually be capital and infrastructure rather than demand for GPUs.
Bank of America Forecasts Semiconductor Market Nearly Doubling to $3.2 Trillion by 2030
Bank of America forecasts the global semiconductor market will nearly double to $3.2 trillion by 2030 from $1.7 trillion in 2026, citing AI infrastructure, memory and data-center demand that remains far stronger than recent fears suggest. Analysts led by Vivek Arya wrote that despite rising concerns about a potential AI infrastructure and investment slowdown, they see no signs of slowing in customer orders, long-term agreements, capacity commitments or semiconductor pricing. The bank pointed to Nvidia B200 rental pricing at $5.72 per hour, up consistently over the past two months and less than 10% below its March peak of roughly $6.10, while AMD has also indicated it is not seeing weakening customer orders. Within the total, memory is expected to remain the biggest growth engine, with sales rising to $1.8 trillion by 2030 from $937 billion in 2026, core semiconductors projected to increase to $1.35 trillion from $739 billion, server-related sales jumping to $848 billion from $359 billion, and wafer-fabrication-equipment spending more than doubling to $359.8 billion from $155.9 billion. BofA sees Nvidia and AMD as resilient compute plays, Marvell in networking, and Analog Devices and onsemi in analog, while Micron, Lam Research, Applied Materials and Intel could also outperform if sector momentum improves; the key risk is whether hyperscalers eventually slow AI capital spending.
Nvidia Leads Q2 Chip Earnings as Group Beats Revenue Estimates by 6.3%
Nvidia reported revenues of $96.22 billion for the second quarter, up 106% year on year and 4.2% above analysts' expectations, as the nine processors and graphics chips stocks tracked in the group collectively beat revenue consensus by 6.3% and guided next-quarter revenue 6.8% above estimates. Intel posted the strongest quarter in the group, with revenues of $16.13 billion, up 25.4% year on year and 11.7% ahead of expectations, though its shares are down 3.2% since reporting and trade at $97.06. Qualcomm was the weakest, with revenues of $9.95 billion, down 4% year on year but still 3% above expectations, while its stock has risen 20.5% since results to $187.57. Lattice Semiconductor reported revenues of $201.1 million, up 62.2% year on year and 8.6% above expectations, and posted the highest guidance raise among its peers, yet its shares are down 23.5% at $105.51. Penguin Solutions reported revenues of $478.7 million, up 47.6% year on year and 17.5% above expectations, the biggest analyst estimate beat of the group, but its stock is down 24.4% at $47.41. On average, share prices across the group are down 3.6% since the latest earnings results, while Nvidia shares are up 1.3% since reporting and trade at $212.38.
Altimeter's Gerstner Blames AI 'Negativity' on Political Agenda, Opens Qualcomm and Micron Positions
Brad Gerstner, founder and CEO of Altimeter Capital, said on CNBC that he supports AI safety measures but believes recent fears are being exaggerated to advance a political agenda, adding that he wants an investigation into who is behind the negativity and where the money is coming from. Altimeter's second-quarter 13F filings show the fund opened new positions in several technology and AI-related companies, including Qualcomm and Micron Technology. Altimeter bought 1.88 million Qualcomm shares valued at about $347.9 million as of June 30, a position accounting for 3.54% of the fund's reported US equity portfolio. It also acquired 209,520 Micron shares worth roughly $241.8 million, representing 2.46% of the portfolio. Bulls argue the market still views Qualcomm as a smartphone-chip company and is overlooking its potential as an AI data-center supplier, pointing to its High Bandwidth Compute architecture, which Qualcomm says delivers 6 times more bandwidth per watt than high-bandwidth memory and which Microsoft and Meta have agreed to use. The main risk is that HBC remains unproven, with the first chips set to arrive with Qualcomm's upcoming AI250 accelerator and no testing yet at scale in real data centers.
Nvidia and SK Hynix Signal Higher Memory Prices Ahead of Micron Results
Micron Technology is set to report its fiscal 2026 fourth-quarter results on Sept. 30, with Nvidia and SK Hynix signaling that the memory price surge driving Micron's growth will continue. Nvidia, reporting its fiscal 2027 second-quarter results last month, said it faces extreme pricing conditions in memory that exceeded prior expectations and are headed even higher into next year, with CFO Colette Kress warning of gross margin pressure from higher component costs. SK Hynix expects the memory shortage to worsen next year and estimates demand could outstrip supply beyond 2030, while KB Securities estimates SK Hynix holds just 10 days of memory inventory. Nvidia estimates capital spending by the top five hyperscalers could rise to $1.3 trillion in 2027 from $800 billion this year. Consensus estimates project Micron's fiscal 2027 revenue rising 88% to $244.5 billion and earnings per share jumping 112% to $156.07, with the stock trading at 6.6 times forward earnings versus 21 for the S&P 500.
AMD's Path to $750 by 2027 Hinges on Helios and MI500 Ramps
Advanced Micro Devices is riding a data-center surge that has lifted its shares 128.8% year-to-date, with Data Center revenue more than doubling to a record $6.7 billion last quarter, while NVIDIA has gained just 13.63% over the same period. Lisa Su has locked in gigawatt-scale commitments from OpenAI, Meta, and Anthropic, with Anthropic alone set to deploy up to 2 GW of MI450 Series GPUs on top of 6 GW each from Meta and OpenAI. Wall Street's consensus target sits at $615.07, but the bull case for $750 by 2027 rests on FY2027 EPS consensus of $15.51, up from $13.10 ninety days ago, which would put the target at just 48 times forward earnings. Su guided Data Center segment revenue to more than double year-over-year in 2027 and said customer pull for Helios is very strong and tracking ahead of initial forecasts, with server CPU revenue expected to grow 70%-plus. The primary risks are any delay to the MI500 ramp or a fresh round of U.S. export controls on the MI308 to China, either of which would reset the model.
Data Center Capex Jumps 92 Percent in 2Q 2026 on AI Demand and Memory Costs
Worldwide data center capital expenditures accelerated sharply in 2Q 2026, growing 92 percent, according to a newly published report from Dell'Oro Group. Continued AI infrastructure investment supported growth across compute, storage, networking, and physical infrastructure, while rising memory and storage prices significantly increased server average selling prices. Baron Fung, Vice President of Research at Dell'Oro Group, said spending remained concentrated in NVIDIA Blackwell Ultra and hyperscaler custom accelerators, while agentic AI created incremental demand for general-purpose compute, storage, and complementary networking. Neocloud providers and AI model builders posted the fastest capex growth among customer segments, reflecting the early stages of their infrastructure buildouts, and Dell led server OEM revenue, followed by Supermicro and Lenovo, while white-box server revenue reached a record high. Fung added that ongoing accelerator deployments and emerging agentic AI and AI-related storage workloads should sustain strong capex growth through the remainder of 2026 and beyond, although supply constraints could limit the pace at which planned infrastructure is deployed.
ASML and TSMC Post Q2 2026 Results as EUV Monopoly Meets Foundry Capex Surge
ASML and Taiwan Semiconductor Manufacturing both reported Q2 2026 results in mid-July, with ASML raising its full-year outlook on AI-driven lithography demand and TSMC lifting its capex plan while pushing 2nm into commercial production. ASML delivered $10.65 billion in revenue, up 21.3% year over year, with operating margin expanding to 37.1%, and memory-related system sales now expected to grow over 75% this year. TSMC's revenue hit $40.20 billion, up 36.0%, with gross margin at 67.7%, as HPC drove 66% of revenue and advanced nodes at 7nm and below made up 77% of wafer revenue. TSMC raised its 2026 capital budget to $60 billion to $64 billion and announced another $100 billion Arizona investment, capital that ASML will help spend. ASML is close to being fully covered with low-NA EUV orders for 2027 and is studying a further 30% capacity expansion for 2028, with CEO Christophe Fouquet saying, "We're not waiting. We're preempting."
Anthropic Commits $517 Billion to Compute, Boosting Nvidia, AMD, Broadcom, Alphabet and Amazon
Anthropic agreed to spend $517 billion on computing power over the 11 months through the end of August, according to The Information, far above the $180 billion in commitments it had previously disclosed. The spending is set to benefit Nvidia, whose GPUs remain the primary chips powering AI workloads and which could invest up to $10 billion in Anthropic as an anchor investor in its IPO. Advanced Micro Devices will invest up to $5 billion in Anthropic, which will buy AMD's Helios rack-scale system and deploy up to 2 gigawatts of AMD GPUs beginning in 2027. Anthropic is also set to become Broadcom's largest customer as it deploys Tensor Processing Units that Alphabet co-designed with Broadcom, with Broadcom saying Anthropic will deploy 1 gigawatt of Alphabet's Ironwood TPUs this year, 5 gigawatts of its TPU v8i in 2027 and another 10 gigawatts in 2028. Amazon, which owns around a 20% stake in Anthropic, has received more than $100 billion in AWS cloud commitments from the lab over the next decade.
US Lacks Effective Tools to Halt China's AI Catch-Up as Rare Earth Retaliation Looms
The United States is struggling to find effective means to prevent China from becoming an AI superpower. Anthropic CEO Dario Amodei, in a 3,800-word essay, called for banning sales of advanced AI semiconductors to China and tightening crackdowns on smuggling, but President Trump has signaled openness to allowing sales of some chips to China, including Nvidia's H200. The biggest challenge for the US is avoiding provoking further tightening of export controls by the Chinese government on rare earths used in precision-guided missiles and drones. Last year, President Trump and President Xi Jinping agreed to a truce in which China guaranteed rare earth supplies in exchange for tariff reductions, and that agreement is likely to be extended when Xi visits the White House next week. Although China still lags in semiconductor production, AI models developed by companies such as DeepSeek and Moonshot AI are narrowing the capability gap with the most advanced US models, often at a fraction of the cost.
Nvidia CEO to Attend Trump-Xi Dinner, AI Talks in Focus
Jensen Huang, CEO of Nvidia, is expected to attend a dinner hosted by President Donald Trump in honor of Chinese President Xi Jinping next week. Reports citing sources familiar with the matter say Huang is expected to attend the event on September 24, during Xi Jinping's visit to Washington for talks with Trump. Huang's attendance comes as the United States and China prepare for negotiations on economic and technological issues, and the meeting between Trump and Xi Jinping is expected to cover artificial intelligence, or AI, as well. According to CNBC, Nvidia is a major chip supplier for AI developers, while the United States and China remain the two leading countries in developing and deploying advanced AI systems, making Nvidia closely watched amid rapid AI development and technological competition between the two countries. Huang has close ties with Trump and was among a group of U.S. business executives selected to travel to China with Trump earlier this year. Huang also previously attended a dinner hosted by Xi Jinping in honor of Trump during the U.S. president's visit to China in May, which was also attended by Elon Musk and Tim Cook, the former CEO of Apple.
Nvidia in Talks to Invest Up to $10 Billion in Anthropic IPO
Nvidia is in advanced talks to invest up to US$10b in Anthropic's planned IPO, according to people familiar with the discussions. The chipmaker is also exploring a potential acquisition of AI platform Hugging Face for up to US$14b, a deal that would expand its software footprint. Both moves would extend Nvidia's role beyond AI hardware into deeper capital partnerships and ownership stakes in key model and tooling providers. Backing a major model provider while already supplying compute would shift Nvidia from supplier to financial partner, while owning a widely used open-source hub could strengthen CUDA-centric workflows and differentiate it further from AMD and Intel. The Hugging Face discussions also sharpen existing risks around regulatory scrutiny, customer vertical integration and energy-intensive AI factory buildouts.
Trump Calls AI a Hoax After Unexpectedly Summoning Nvidia CEO Jensen Huang On Stage
President Donald Trump unexpectedly called Nvidia CEO Jensen Huang up on stage and called AI a hoax, according to Seeking Alpha. The remark lands as AI companies spend hundreds of billions of dollars on chips, data centers and power infrastructure, and as industry leaders debate whether research should stall. Nvidia has emerged as one of the greatest financial winners of the AI buildout, with its accelerators driving many of the biggest computing systems in the industry, so any significant slowdown in investment would quickly draw investor notice. Governments are already squabbling over AI legislation, energy needs, semiconductor exports and national security limits, putting Nvidia at the center of many of those disputes. While a single statement does not affect the company's huge order pipeline, investors will now watch whether the political discussion remains hyperbole or starts to alter the policies and spending that support Nvidia's growth.
Meta to Boost In-House AI Chip Use to Cut Infrastructure Costs
Meta Platforms is planning to increase its use of in-house-designed AI chips as it looks to reduce the cost of its vast infrastructure buildout. The company has invested billions in data centers, servers and accelerators as it builds more sophisticated models and rolls out AI features across Facebook, Instagram and WhatsApp. Custom silicon can be tuned around particular workloads rather than the wide range of tasks expected from general-purpose accelerators, making AI services cheaper and more efficient to run at massive scale. More internal chips would lower Meta's reliance on outside semiconductor providers for some tasks, even as sophisticated Nvidia accelerators remain core to its infrastructure. The move follows a wider trend among hyperscalers such as Google and Amazon that have developed specialized processors for AI and cloud computing, and Meta's next test is whether those chips can deliver considerable savings as its AI infrastructure spending keeps growing.
Jim Cramer Turns More Bullish on Oracle After 121% Cloud Infrastructure Revenue Jump
Jim Cramer said Oracle Corporation's latest quarter left him more bullish on the company's artificial intelligence buildout, pointing to a 121% jump in cloud infrastructure revenue and a $664 billion remaining performance obligation as evidence that AI demand is translating into contracted business. Oracle's fiscal first-quarter 2027 revenue rose 30% year over year to $19.3 billion, while total cloud revenue increased 62% to $11.61 billion and cloud infrastructure revenue reached $7.4 billion. The company delivered 850 megawatts of additional data-center capacity and more than 300,000 GPUs, and its RPO increased $209 billion from a year earlier to $664 billion, with management expecting roughly half of that backlog to convert into revenue over the next 36 months. Oracle reported $28.499 billion of capital expenditures but received $11.363 billion of customer prepayments with a significant financing component, putting net cash outlay for CapEx at $17.966 billion, and it completed its previously announced $20 billion at-the-market equity offering. The company expects fiscal 2027 revenue of at least $90 billion and non-GAAP EPS of $8.10, with current-quarter revenue growth of 30% to 34% and cloud-revenue growth of 65% to 71%, though the stock closed September 11 down 1.74% at $150.28 as investors weighed negative free cash flow of $5.4 billion and $90 billion to $95 billion of expected gross CapEx for fiscal 2027.
ASML Studies Building Over 110 EUV Systems in 2028 as 2027 Capacity Fills
ASML Holding is studying ways to manufacture more than 110 extreme-ultraviolet lithography systems in 2028 after its 2027 capacity became nearly fully committed, according to Reuters. The company expects to have capacity for at least 80 EUV systems in 2027, so expanding to more than 110 would represent an increase of at least 37.5%. ASML's second-quarter results showed 9.3 billion in sales and a 54% gross margin, while management lifted full-year revenue guidance to 43 billion to 45 billion. The shares gained approximately 0.5% to $1,582.91 in U.S. trading Tuesday, a level that sits 25.29% above a GF Value of roughly $1,260. With High-NA EUV systems carrying prices around $400 million, even a handful of extra deliveries can move revenue materially.
Dell Shares Rise 5.8% as Axelera AI's Europa Chip Enters Its Servers
Dell Technologies shares rose about 5.8% to $565.15 on Tuesday after European startup Axelera AI unveiled its Europa inference processor, which Reuters reported is designed for enterprise inference workloads and is expected to appear in future systems from Dell and Supermicro. Axelera says more than 600 customers already use its technology, with signed agreements worth tens of millions of dollars and a broader potential sales pipeline that could reach $1.5 billion, an opportunity estimate rather than booked revenue. Dell's larger AI engine remains its server business, which exited the latest quarter with roughly $95 billion of AI-server backlog after shipping $16.4 billion of AI systems. Even if Axelera captured the full $1.5 billion opportunity, that would equal only about 1.6% of Dell's existing AI backlog. The strategic takeaway is that Europa gives Dell another inference supplier and potentially more power-efficient hardware choices, though Nvidia-based systems still carry far more weight in Dell's near-term AI economics.
Wells Fargo Cuts S&P 500 Year-End Target to 7,700, Warns of 10% Pullback
Wells Fargo lowered its year-end forecast for the S&P 500 to 7,700 from 7,950, warning that stocks could face a five percent to 10% pullback before year-end as earnings momentum matures. The bank raised its earnings estimates to $425 for 2027 and $460 for 2028, but analyst Ohsung Kwon expects weaker valuation multiples to offset part of that projected profit growth. Kwon estimates equities make up 72% of portfolios, the highest share since 1969, and with the 10-year Treasury yield approaching 5%, Wells Fargo sees about 60% as a more appropriate equity allocation, leaving a 12-percentage-point gap. The bank also shifted its sector preferences, moving Technology to Equal Weight from Overweight and raising Health Care to Overweight from Equal Weight. Within Technology, Kwon favors Software over Semiconductors, citing political resistance to data-center projects and currency pressure involving the Korean won as risks for chip companies, and Wells Fargo expects semiconductor stocks could revisit their July lows.
Crusoe and Perplexity Sign Multi-Year AI Infrastructure Partnership
Crusoe announced a multi-year partnership with Perplexity to power Perplexity's AI workloads on Crusoe Cloud, spanning the full model lifecycle from frontier model training to production inference. Under the agreement, Perplexity will train frontier models on Crusoe's dedicated NVIDIA GB300 NVL72-powered clusters and NVIDIA InfiniBand Solutions, and serve them in production through Crusoe's Managed Inference service. In return, Crusoe will adopt Perplexity Enterprise Pro and Max for its 1,800 employees, giving them a platform for web and internal knowledge search, multi-step research, data analysis, and access to frontier AI models. Crusoe co-founder and CEO Chase Lochmiller said the fastest-moving AI companies need infrastructure that keeps pace across the entire model lifecycle, while Perplexity co-founder and CEO Aravind Srinivas said every millisecond of latency is felt by users at Perplexity's scale. Dion Harris, Senior Director of HPC and AI Infrastructure Solutions at NVIDIA, said Crusoe Cloud enables Perplexity to train, fine-tune, and serve frontier models in production with the speed and efficiency agentic AI demands.
Piper Sandler Initiates AMD at Overweight With $600 Target
Piper Sandler analyst David O'Connor initiated coverage on Advanced Micro Devices with an Overweight rating and a $600 price target on September 9. O'Connor estimates AMD's revenue will grow at a 50% CAGR from fiscal 2026 through fiscal 2030, while earnings per share compound at 65%, reaching $53 by 2030. The $600 target is based on a 24x multiple of fiscal 2028 estimated earnings. The bull case rests on agentic AI lifting server CPU demand, with AMD's Venice portfolio and its Helios GPU ramp, which has signed anchor clients in OpenAI, Meta, and Anthropic. AMD reported second-quarter revenue of $11.5 billion, including $6.7 billion from data centers, up 107% year-over-year, though the year-ago quarter included an $800 million export-related inventory charge tied to U.S. restrictions on shipping the Instinct MI308 AI accelerator to China, which overstated the year-over-year margin comparison. Hedge fund sentiment strengthened in the second quarter of 2026, with 164 hedge funds holding stakes, up from 134 in the prior quarter.
Piper Sandler Starts Broadcom at Overweight With $460 Target
Piper Sandler initiated coverage of Broadcom Inc. with an Overweight rating and a $460 price target on September 9, with analyst David O'Connor calling the company the leader in ASIC chips with a 75% share of the ASIC market for AI inference. O'Connor estimates demand is running twice the available supply, making component availability and deployment timing the key determinants of whether that demand converts into earnings. Broadcom is already ramping Meta's MTIA program and OpenAI's Jalapeno chips along with two other customers, while Google's TPU program remains its major volume ASIC program and Anthropic scales deployments on AVGO-enabled TPU compute; a networking attach rate of around 30% could generate an estimated $20-30 billion of total content per gigawatt of deployed capacity. Piper sees a line of sight to 12 gigawatts of demand in fiscal 2027, rising toward an estimated 38 gigawatts by fiscal 2030, underpinning roughly 51% EPS compound growth through the decade, and its $460 target is based on a 14x multiple of estimated fiscal 2028 earnings. The firm also flags risks including eroding pricing power, customer concentration and lower relative gross margins, while 170 hedge funds held the stock in the second quarter, down slightly from 173 in the prior one.
Analyst: AI Chip Sell-Off Overblown, No GPU Slowdown After Anthropic Essay
Lynx Equities Strategies analyst KC Rajkumar says the AI semiconductor sell-off triggered by Anthropic CEO Dario Amodei's essay on slowing AI capability development is a misreading of his actual message, with zero supply-chain evidence of any hardware pullback. In a note titled "AI Semis: Tempest in a teacup," Rajkumar said his supply-chain checks found no slowdown in GPU rollout, no cancellations in memory and flash, no cancellations at TSMC and OSAT companies, no changes to delivery schedules at semicap companies, and no disruption in the physical components ecosystem. He pointed to the essay itself as exculpatory, writing that "nowhere in the Dario essay is there a call to slow down the pace of hardware roll-out," and argued Anthropic has a direct financial incentive to keep compute capacity expanding because halting technical progress would impact revenue growth and its path to profitability and IPO. Rajkumar attributed the real weakness in AI semiconductor names to macro headwinds from geopolitics and the long bond, including US-China chip export control tensions and Taiwan-related risk, rather than any shift in AI industry sentiment. He sees a potential mini bounce in AI semis and hardware stocks ahead of the FOMC event, with the sector more leveraged to whether the long bond yield flattens out, and said the widely anticipated quarter-point hike, expected at the September 16-17, 2026 FOMC decision, is already priced in.
Analysts See Up to 136% Upside in Nvidia and Micron
Select Wall Street analysts see up to 136% upside in two trillion-dollar AI infrastructure stocks, Nvidia and Micron Technology, based on their Sept. 11 closing prices. Raymond James analyst Simon Leopold expects Nvidia to reach $515 per share, implying 136% upside, citing triple-digit Data Center sales growth, GPU supply shortages that could ease, and the potential for non-hyperscaler sales growth to outpace hyperscaler growth. Melius Research's Ben Reitzes believes Micron can soar to $2,200 per share, representing 126% upside, after the company announced it had secured at least $100 billion in strategic customer agreements through 2030 for its memory solutions. Reitzes has raised his firm's price target twice since late April, pointing to off-the-charts demand for high-bandwidth memory and persistent supply shortages for HBM, DRAM, and NAND flash that should sustain Micron's pricing power and historically elevated margins. As of the article's writing, 58 of the 61 covering analysts had a buy- or strong buy-equivalent rating on Nvidia.