Amazon.com Inc

Nvidia, the major U.S. AI chip maker, announced its latest quarterly results (May-July 2026) with revenue reaching $96.2 billion, up 106% year-over-year and above analyst expectations of about $92 billion. Net income surged to $59.7 billion, or $2.46 per share, up 126% from the same period last year. The data center business, a key revenue driver, brought in $89 billion, more than doubling. Meanwhile, the edge computing business generated $7.2 billion, up 27%. Nvidia forecasts current-quarter revenue (August-October 2026) of $108 billion, above market expectations, and expects next fiscal year's revenue to grow another 70%, citing supply chain constraints as a limiting factor on sales, not weakening demand. Major customers such as Amazon, Microsoft, Alphabet, and Meta are expected to invest nearly $800 billion this year to develop data centers and AI infrastructure.

Bath & Body Works reported second-quarter net sales of $1.5 billion, down 2.3% year over year but beating its guidance range of down 5% to down 3%, and adjusted earnings per diluted share of $0.62, well above the $0.20 to $0.25 range, including a benefit from about $80 million in tariff refunds. The company raised its full-year adjusted EPS guidance to $2.60 to $2.80 and narrowed net sales guidance to down 4% to down 2.5%, while increasing free cash flow expectations to approximately $650 million. Direct channel sales grew 3% to $275 million, and international and other net sales rose 24.9% to $108 million, but U.S. and Canadian store sales fell 5.4% to $1.1 billion. CEO Daniel Heaf noted progress in digital growth and expanded distribution, with Amazon net sales more than tripling sequentially and Ulta Beauty launching in about 600 stores, but acknowledged continued store traffic pressure and a weak third-quarter outlook with adjusted EPS forecast at $0.07 to $0.12.

Nvidia reported strong fiscal second-quarter 2027 results that beat expectations, with adjusted earnings per share of $2.22, above the analyst estimate of $2.10, and revenue of $96.22 billion, above the expected $92.17 billion, reflecting still-hot demand for AI chips. The stock rose about 4% after the earnings release. The company expects revenue in fiscal 2028 to grow as much as 70%, higher than the market's forecast of 44%. The CFO noted that supply constraints are still a consideration. For the current quarter, Nvidia expects revenue of about $108 billion, plus or minus 2%, above the analyst estimate of $104.2 billion. The data center business remains the main engine, generating $89 billion in revenue, up 117% year over year, accounting for about 92% of total revenue. Meanwhile, Amazon Web Services plans to purchase 2 million GPUs and adopt the new "Vera" CPU, reflecting that hyperscalers continue to invest heavily in AI. However, Nvidia faces pressure from higher memory costs and increased debt, with $33.5 billion in outstanding bonds and a $25 billion commercial paper program, while also returning $26 billion to shareholders in the latest quarter.

Einride posted a 27% increase in revenue for the first half of 2025 following its June 10 IPO, and the company expects that growth rate to roughly double in the second half of the year, CEO Roozbeh Charli said in an interview with FreightWaves. The results represent early progress toward converting approximately $800 million in joint business plans with customers into recognized revenue. Alongside the revenue gains, Einride grew its autonomous driverless operating hours by about 60% in the first half, with roughly 250 trucks deployed on its Saga AI platform across more than 30 customers in seven countries, including Amazon, Heineken, and PepsiCo. The most significant near-term expansion is a 500-truck order for Tesla Semi vehicles, which Einride will purchase, finance through third-party partners, and deploy on its platform as the carrier of record; approximately 75 of those trucks are expected to be deployed before year-end, with the remainder targeted for deployment by the end of next year. Separately, Amazon's middle mile freight network is absorbing 75 Einride electric heavy-duty trucks across five locations, a deal announced in April before the IPO, with the majority expected to be deployed before year-end. Einride is also pursuing a defense business unit built around its Android Driver autonomous software stack, having conducted a pilot for a NATO allied country and added retired General Keith Alexander to its board.

Amazon.com Inc. will add an additional 2 million Nvidia Corp. graphics processing units to its data center fleet in the next two years, a sign that the company remains committed to the AI hardware leader's products despite its own competitive chipmaking efforts. The two companies said on Wednesday that Amazon would deploy some 2 million of Nvidia's high-end chips, which are designed to train and run artificial intelligence models, in 2027 and 2028. That's in addition to the 1 million Nvidia chips Amazon in March said that it would be installing in Amazon Web Services data centers, beginning this year. The chips in the latest deal include Nvidia Blackwell Ultra, Rubin and Rubin Ultra GPUs. Those products will likely carry a list price of at least tens of thousands of dollars a unit, though big customers can get discounts. Amazon and Nvidia didn't announce financial terms of their latest deal.

Amazon Web Services and NVIDIA announced a major expansion of their strategic collaboration, planning to deploy 2 million additional NVIDIA GPUs across AWS's global infrastructure in 2027-2028. This new commitment builds on earlier plans announced at NVIDIA GTC 2026 to add more than 1 million GPUs starting in 2026, reflecting surging demand for AI infrastructure. The expanded partnership also includes bringing NVIDIA Vera CPU-based infrastructure to AWS, extending NVIDIA NVLink Fusion with custom high-bandwidth memory, and building AI factories for the U.S. government with 100,000 GPUs for federal and national-security workloads. Additionally, AWS will integrate the NVIDIA platform with its Nitro System and Elastic Fabric Adapter, support NVIDIA Nemotron open models on Amazon Bedrock and SageMaker, and accelerate data processing and robotics workloads. AWS CEO Matt Garman and NVIDIA CEO Jensen Huang both emphasized the unprecedented scale and pace of AI adoption, with Huang noting that demand is running ahead of every forecast.

Amazon has acquired Amsterdam-based software developer DuckLabs to strengthen its AWS analytics offerings, though financial terms were not disclosed. The deal does not include DuckDB, DuckLake, or Quack, which remain separate open-source projects under the nonprofit DuckDB Foundation. Founders Hannes Muhleisen and Mark Raasveldt will stay on to lead the team and technical work. Amazon expects the acquisition to make AWS analytics products easier to use and more economical, as cloud providers compete on data processing tools. The move is unlikely to materially affect Amazon shares in the near term but could bolster its broader cloud strategy.

TRUE has announced a strategic partnership with AWS to enhance its use of AI and study the deployment of Amazon's low-orbit satellite connectivity service, Leo, to expand broadband coverage in Thailand. After completing its structural cost cleanup, the company is now focusing on driving revenue growth. In the second half of the year, the B2C business continues to show steady growth, while B2B is slowing due to economic conditions. Data usage has increased 21% year-on-year. TRUE continues to maintain strict cost control and is leveraging AI and automation technologies to improve efficiency. The partnership with AWS covers cloud management, AI-powered customer service, and predictive network maintenance. The Leo satellite will serve as a complementary service alongside terrestrial networks and WiFi. Currently, TRUE is in discussions with Amazon Leo to explore ways to offer the service in Thailand.

Amazon has expanded its delivery methods by using Japan's Shinkansen bullet trains for line-haul transportation and deploying e-cargo bikes in the U.S. for the first time. The trains, which travel up to 200 mph, enable same-day delivery from Tokyo to parts of central and northern Japan, reducing transit time and carbon emissions. In the U.S., Amazon launched e-cargo bike operations in Culver City, California, with 25 bikes, and plans to expand to Burbank. These bikes, which require a drivers' license to operate, open a wider pool of hiring candidates and can carry around 240 packages per shift. Amazon is also scaling up drone delivery and uses boats for deliveries to remote locations like Monhegan Island.

Moor Insights & Strategy founder Patrick Moorhead predicts Nvidia's AI data center market share could fall from 90% to roughly 70% over the next two years, as the market more than doubles annually. Moorhead identifies Google's TPU as the top taker of unit share, benefiting companies like Broadcom and Marvell, followed by Amazon's Trainium, which underpinned Anthropic's model development. He also expects AMD to gain unit share with its new scale-up architecture, which is more similar to Nvidia's than previous designs. Regarding Bloomberg's report that OpenAI's custom chips with Broadcom outperformed Nvidia's current lineup in testing, Moorhead says he is not surprised, noting the power and performance of non-GPU architectures, but he emphasizes that GPUs offer more flexibility and longevity, while custom designs require frequent updates and cost between 200 and 500 million dollars each. Despite these challenges, Moorhead believes Nvidia's revenue won't be significantly impacted because the market is growing so fast, and in the long term, it will remain primarily a GPU world with custom inference chips for those who roll their own software.

AI assistants are becoming a new gateway to retail, changing how consumers discover, compare, and buy products, according to a report from Retail Insight Network. McKinsey research cited in the report found that 38% of surveyed consumers in France, Germany, and the UK use AI to research products or inform purchase decisions. Major platforms are moving in this direction: Amazon renamed its Rufus assistant to Alexa for Shopping in May 2026, and OpenAI launched shopping research in ChatGPT in November 2025, with retailers like Target, Sephora, and Nordstrom integrating via its Agentic Commerce Protocol. McKinsey estimates agentic commerce could mediate between $3 trillion and $5 trillion of global consumer commerce by 2030. The report advises retailers to audit product data, monitor AI visibility, prepare for AI integration, and protect direct customer relationships as AI becomes another front door to retail.

Charoen Pokphand Group (CP Group), AIS Ventures Group (AIS), True Corporation, and Amazon have announced a five-year strategic partnership to propel CP Group and AIS into a regional AI Lighthouse role, while enhancing AI capabilities and strengthening Thailand's digital economy. The collaboration covers five key dimensions: developing digital and AI skills for 3 million Thais, elevating True into an AI-First Organization, using AWS in Ascend Money's financial services, establishing an AI Center of Excellence, and transforming over 19,000 service points of 7-Eleven, Lotus's, and Makro under the Thailand AI National Mission framework, with the goal of making Thailand a regional AI hub.

Chinese AI startup Moonshot AI is in talks to provide its popular model Kimi K3 to major US cloud providers in exchange for a share of revenue from related services, according to three people familiar with the matter. The discussions involve Microsoft, Amazon.com, and Google, and if a deal is reached, it could mark the first major revenue-sharing agreement between a Chinese AI company and a US cloud provider. Moonshot is seeking up to 30% of revenue from Kimi K3-related services offered on Microsoft's Azure, Amazon Web Services (AWS), and Google Cloud. This is said to be in line with the terms offered to large customers using Kimi K3. The talks are at an early stage, and there is no guarantee a deal will be reached. The US government has banned exports of AI semiconductors to China over national security concerns, and senior officials have criticized Moonshot, but powerful Chinese AI models are often significantly cheaper than Western counterparts, and their use is growing in the US. Unresolved issues in the discussions include how revenue will be split, access to data, and auditing of token usage.

Anthropic, the AI lab founded in 2021 by former OpenAI employees, is set to become the most anticipated IPO of 2026 after filing its Form S-1 with the SEC on June 1st, with Polymarket betting showing an 83% chance of a public debut by the end of October and 93% by year-end. With a private valuation of approximately $965 billion and an annualized revenue run rate of about $65 billion, Anthropic is the fastest-growing technology company in history, specializing in enterprise AI and software coding through its Claude models. Several public companies hold significant stakes, including Amazon, whose $43 billion stake is only about 2% of its market cap, while SK Telecom's $2.7 billion stake represents roughly 20% of its market cap, making it the biggest relative beneficiary. The IPO is seen as a crucial barometer for the AI industry, given that AI spending accounts for the lion's share of US GDP growth.

Amazon's investment narrative has shifted from retail efficiency to aggressive AI infrastructure spending, with management planning approximately $200 billion in cash capital expenditures in 2026, mostly for AWS and AI. AWS is now a $169 billion annualized revenue run rate business, with chips and AI-specific revenue each exceeding $25 billion, and AWS grew 37% year-over-year in Q2 2026. AWS generated $16.7 billion, or over 60%, of Amazon's $27.5 billion total operating profit in Q2 2026, while North America retail remains a $426.3 billion annual segment growing 10%. The shift signals that holding Amazon stock is now fundamentally a bet on high-margin AI infrastructure rather than retail volume.

Growing political and community pushback is creating new headwinds for data center stocks, with concerns ranging from rising electricity demand to water consumption. Pennsylvania, Texas and Michigan have all moved toward tighter oversight or restrictions on new projects, and generous tax incentives once used to attract data center investments are increasingly facing political scrutiny. Jim Cramer of CNBC sees the political backlash as a major advantage for hyperscalers such as Amazon, Alphabet, Microsoft and Meta, whose financial strength could allow them to continue developing data centers even as smaller, speculative developers face greater hurdles. Investors seeking concentrated hyperscaler exposure can consider the Defiance AI Hyperscale Leaders ETF, the Roundhill Magnificent Seven ETF, and the Corgi Mag 7 ETF.

Top investors on The Investing for Beginners Podcast named Taiwan Semiconductor Manufacturing, ASML, Amazon Web Services, and Microsoft Azure as AI stocks that could win regardless of which model leads. TSMC ranks first for its manufacturing moat, guiding slightly above 40% year-over-year 2026 revenue growth in U.S. dollar terms, while ASML's EUV lithography monopoly is nearly fully booked for 2027. AWS delivered its fastest growth in 18 quarters at 36.7%, while Azure grew 43% backed by $678 billion in commercial backlog. Frontier labs OpenAI and Anthropic land in the too-hard pile, being private companies whose competitive edge hinges on winning an unpredictable model race.

Jim Cramer argued on Mad Money that political resistance to data center construction may strengthen Microsoft, Alphabet, and Amazon by eliminating speculative builders and easing cost pressures. He cited Texas Governor Greg Abbott and Pennsylvania Governor Josh Shapiro shifting from boosters to critics, and said the hyperscalers can afford to compensate communities while spec builders get obliterated. Cramer also warned the AI trade is currently broken and price-to-earnings multiples are unlikely to expand, advising investors to sell some holdings. Microsoft guided fiscal 2026 capital expenditures to $115.95 billion, Alphabet's Q2 capex reached $44.92 billion, and Amazon spent $54.21 billion in Q2 with an AWS backlog of $496 billion.

SandboxAQ announced that AQCat, its Large Quantitative Model for catalyst discovery, is now generally available in AWS Marketplace and deployable through Amazon SageMaker. The release makes the tool available on demand to materials innovators and industrial teams, who can run it directly inside their own AWS environment without specialized computational scientists or custom code. AQCat was trained on 13.5 million quantum chemistry calculations spanning 47,000 catalyst systems and predicts key catalyst properties at up to 20,000 times the speed of conventional physics-based methods while approaching their accuracy. The launch extends a cloud go-to-market strategy that also spans Anthropic, where SandboxAQ offers access to AQCat via MCP. AQCat is generally available to AWS commercial customers in North America.

Synnex Thailand Public Company Limited, or SYNNEX, has launched the AI Hub project in Thailand, built on AWS cloud and artificial intelligence infrastructure to support Thai organizations in applying AI for business purposes. The project aims to be a hub connecting world-class AI and cloud technology with local expertise, partner networks, and solutions that meet real-world needs, enabling organizations from SMEs to large enterprises to access, trial, and scale AI more quickly. The AI Hub will be driven by three main mechanisms: increasing access to AWS AI and cloud technology, leveraging partner networks to support from solution development to market launch, and integrating global technology with expertise that understands the Thai business context. Currently, SYNNEX has applied AI internally in nearly 90 use cases, reporting a 55% increase in work efficiency, a 59% reduction in work time, and a 60% reduction in errors, while 94% of employees use AI at least once a week.

Drone delivery is moving from pilot projects toward a credible role in retail fulfilment, with Amazon planning to expand Prime Air to nearly 500 US cities and towns by the end of 2026 and Walmart having completed more than one million drone deliveries. Walmart is working with Wing on a network of more than 270 locations planned for 2027, while DoorDash received FAA Part 135 air-carrier certification in July and launched DoorDash Air, and Uber has agreed to introduce Zipline drone delivery through Uber Eats with a target of one million drone deliveries a day by the end of 2029. The future is likely a hybrid model in which drones handle small, lightweight, time-sensitive purchases in selected suburban and metropolitan areas, while vans and couriers continue to serve larger or consolidated orders. Regulation remains a key gateway, with the FAA proposing a Beyond Visual Line of Sight framework in 2025 and the UK Civil Aviation Authority aiming to enable routine BVLOS operations through 2027, but cost competitiveness will depend heavily on utilisation, delivery density, and increasing autonomy.

Data centers are driving the largest capital investment in US history, with the five biggest hyperscalers—Amazon, Microsoft, Google, Meta and Oracle—set to spend more than $750 billion on capital expenditures this year, up 67% from last year and roughly 75% earmarked for AI infrastructure. The buildout is straining the electric grid, with utilities now forecasting a sixfold jump in 2030 peak demand growth from predictions just three years ago, and data center electricity use is forecast to rise from 4.4% of US consumption in 2023 to almost 12% by 2030. Politically, more than 70% of Americans oppose a data center in their area, and at least 75 projects totaling roughly $130 billion in potential investment were delayed in Q1 2026, with Pennsylvania Governor Josh Shapiro imposing new requirements, Michigan Senate nominee Mike Rogers backing a one-year moratorium, and Texas Governor Greg Abbott freezing new data centers. Meta is committing more than $50 billion to build Hyperion, a 5-gigawatt data center in Louisiana powered by 10 new natural gas plants, while Nvidia announced over $100 billion in guarantees for an 8-gigawatt project in Ohio to power OpenAI.

SYNEX has expressed confidence that total revenue this year will reach its target of 53 billion baht, supported by the AI trend and digital transformation investment from both the public and private sectors, as well as new product launches from key brands such as Apple during the high season in the second half of the year. Ms. Suthida Mongkolsuthi, Chief Executive Officer of Synnex Thailand Public Company Limited, said that second-half operating results will grow better than the first half due to demand for hardware upgrades to support AI and new smartphone releases. Most recently, the company has partnered with Amazon Web Services Thailand Company Limited, or AWS, to launch an AI Hub in Thailand under the concept of Empowering Thailand's AI Transformation, driven by three key mechanisms: AI Accessibility, Market Acceleration, and Local Expertise. It is also laying the foundation for growth through five main pillars covering Partner AI Solutions, Partner AI Delivery Capability, Internal Adoption of AI, Funding for AI Competency Partners, and Technology Showcase. SYNNEX stated that applying AI internally across nearly 90 use cases has helped increase productivity by up to 55 percent and reduce errors by 60 percent. Mr. Watson Thirapatpong, Country Manager of AWS for Thailand, added that this collaboration will help Thai organizations scale their AI adoption efficiently.

Amazon shares rose about 1.8% to $263.16 Monday morning even as the Nasdaq slipped, as investors rewarded the company's accelerating cloud growth. Second-quarter sales surged 20% to $200.6 billion, while operating income hit $27.5 billion. AWS revenue exploded 37% to $42.2 billion, its fastest growth in more than four years. Amazon now expects roughly $220 billion in 2026 capital spending, $20 billion above its previous plan. The stock trades 6.73% above its $246.57 GF Value estimate.

Amazon is rapidly expanding its Prime Air drone delivery program to seven states — Arizona, Florida, Kansas, Louisiana, Michigan, Nebraska, and Texas — with additional metro areas including Chicago, Syracuse, Cleveland, Atlanta, and Boise announced as coming soon. The expansion comes as the FAA is expected to finalize standardized beyond-visual-line-of-sight rules within weeks, a regulatory milestone that could allow operators to scale drone delivery far more broadly than is currently possible under individual waivers. Amazon's current MK30 drone operates at altitudes of roughly 200 to 300 feet and has a delivery range of approximately 7.5 miles from a fulfillment center, with orders under $50 carrying a $3 drone-delivery fee. FreightWaves Parcel & Air Cargo Editor Eric Kulisch said the pending rules could open the floodgates for expansion, while noting that Memphis, Tennessee remains delayed by negotiations with local authorities. Morgan Stanley projected in 2021 that the broader urban air mobility market would reach roughly $1 trillion by 2040 and $9 trillion by 2050 in its base case.

Pinterest has committed $4 billion to Amazon Web Services through 2031, expanding its AI infrastructure and cloud spending. The agreement, announced June 4, supports AI development, search and shopping improvements, and modernization of Pinterest's visual-discovery platform. Pinterest processes over 80 billion searches monthly and its Taste Graph draws on more than 16 billion boards. Second-quarter non-GAAP cost of revenue rose 25% to $245 million, driven by GPU capacity and tvScientific, while non-GAAP operating expenses increased 13% to $629 million. The company expects third-quarter non-GAAP cost of revenue to be roughly flat sequentially due to contractual benefits from the new infrastructure agreement.

Marvell Technology may have a much larger AI opportunity ahead than the market currently recognizes, according to GF Securities. The firm estimates that CXL and DPU silicon will each be worth about $3 billion in 2028, with another $3 billion from MPU solutions for AI token production, and a single customer representing a multi-billion-dollar opportunity. Marvell is also expected to join Amazon's Trainium 3 program, generating around $1 billion in 2026 and $2 billion in 2027. GF Securities further noted that Marvell's optical networking position could help it capture a larger share of future Trainium 4 revenue, and that its equity tie-up with Google may improve its chances of winning future TPU contracts, particularly TPU v10.

Amazon-owned Zoox has begun deploying its fully autonomous vehicles onto the streets of San Francisco, putting the business in more direct confrontation with Alphabet's Waymo. Zoox's vehicles have no steering wheel or pedals, are bi-directional, and seat four people facing each other. The company has clocked over 3 million autonomous miles and carried close to 1 million passengers, and began charging for rides in Las Vegas on Aug. 10 after gaining a government exemption to operate commercially without conventional driver controls. Amazon acquired Zoox for roughly $1.2 billion in 2020.

Tesla will unveil the production Cybercab at a launch event in Austin, Texas, on September 3rd, as the company ramps up its self-driving efforts. The two-seat, purpose-built robotaxi has no steering wheel or pedals and is Tesla's first vehicle designed purely for autonomy, meant to run on its Full Self-Driving software as part of the Robotaxi fleet launched in Austin last year. The launch follows last week's unanimous approval by the Nevada Transportation Authority of permits for Tesla, Waymo, and Uber to run commercial robotaxis in Clark County, authorizing up to 8,000 driverless vehicles over the next 12 months. Tesla drew the largest allocation of around 5,000 robotaxis, though its Cybercab chief engineer told regulators the company expects to field around 2,500 within the year, calling the 5,000 figure a ceiling. Alphabet's Waymo was cleared for up to 1,000 vehicles, while Uber secured 1,000 more through partnerships with Hyundai-backed Motional and Amazon's Zoox, which holds a separate permit for 100.

TD Cowen projects Amazon Web Services revenue will hit $222 billion by 2027, 11% above Wall Street consensus, driven by generative AI demand and Amazon's planned $200 billion in 2026 capital expenditures. The firm also raised its 2026 AWS estimate to $165 billion, 3% above consensus, citing AWS's 24% year-over-year growth to a $142 billion annualized run rate in Q4 2025. Amazon's custom Trainium and Graviton chips now exceed a $10 billion annualized revenue run rate with triple-digit growth, and over 100,000 companies use Amazon Bedrock, with CEO Andy Jassy saying the majority of enterprise AI production demand is still ahead. The main risk is a 37.12% year-over-year decline in free cash flow in Q4 2025 as capital spending surged.

Amazon has raised its 2026 capital expenditure target to $220 billion, $20 billion more than its prior plan, citing higher memory costs. The announcement came alongside second-quarter results showing 20% year-over-year revenue growth, driven by Amazon Web Services' highest growth rate in more than four years. Operating income rose 43.2% to $27.5 billion, with AWS operating income jumping from $10.2 billion to $16.6 billion. CEO Andy Jassy said the AI and chips businesses have each surpassed $25 billion annual revenue run rates. Amazon, Microsoft, and Alphabet control more than 60% of the cloud computing market, and the increased spending is expected to widen the gap with competitors.

Nvidia Corp. is facing increasing competition from rivals, startups, and its own biggest customers as the AI chip market is poised to cross $1 trillion in revenue. Amazon.com Inc. CEO Andy Jassy said in July that if Amazon's semiconductor operation were a standalone business, it would generate annual revenue of more than $25 billion, with sales growing by a percentage in the triple digits. Advanced Micro Devices Inc. saw its data center revenue more than double last quarter to $6.7 billion, while Broadcom Inc. expects to sell $56 billion of AI chips this year. Meanwhile, 150 companies are working on more than 200 different AI semiconductor designs, according to Jon Peddie Research, and AI labs like OpenAI and Anthropic are developing their own chips. Nvidia's earnings report on Wednesday is expected to show revenue nearly doubled last quarter to $92 billion, but some customers have been told that prices of servers containing its AI chips are going up more than 15%.

Peter Thiel's Thiel Macro LLC disclosed a $418.7 million portfolio across eight positions as of June 30, 2026, with seven of the eight bets focused on the physical energy bottleneck constraining AI buildout. The largest disclosed positions are Amazon at $118.0 million, Vista Energy at $75.9 million, Vistra at $59.1 million, American Electric Power at $42.2 million, DTE Energy at $40.3 million, FirstEnergy at $39.9 million, CMS Energy at $39.6 million, and X-Energy at $3.7 million. The filing reflects long US equity positions only and excludes shorts, derivatives, private stakes, and non-US listings. The thesis centers on power supply rather than chips, with utilities like American Electric Power disclosing 69 gigawatts of contracted load additions through 2030 and a $78 billion five-year capital plan, while Vistra leverages its nuclear fleet and the pending Cogentrix deal. Amazon's AWS revenue surged 36.7% year-over-year to $42.2 billion in Q2, and its roughly $500 million investment in X-Energy positions it as both a creator and solver of the AI power bottleneck.

Anthropic, the owner of the Claude chatbot, has an annualized revenue run rate of $65 billion, about seven times what it was at the end of last year. The company has filed with the SEC to go public later this year in an IPO that could value it at $2 trillion or more. Retail investors can gain indirect exposure through companies holding pre-IPO stakes: Amazon with a 21% stake from its $33 billion investment, Alphabet with 15%, Salesforce with $5 billion, and Zoom Communications with $1.3 billion. Alphabet and Amazon previously booked large gains from stakes in SpaceX and Anthropic, respectively, and could see further revaluation if the IPO succeeds.