AI Summary

Market Overview

• AGIX slipped 0.12% DTD, underperforming QQQ (+0.46%), the S&P 500 (+0.13%), and the Dow (+0.45%). However, AGIX's WTD gain of 3.01% and YTD return of 35.91% continue to outpace all major benchmarks.

• Among peer AI ETFs, the strongest DTD performers were AIS (+4.29%), SMH (+4.01%), and FDTX (+3.89%), reflecting broad hardware and infrastructure strength. AGIX's DTD lagged due to weakness in its large-cap application and infrastructure holdings.

• Top DTD positive contributors within AGIX were Astera Labs (+11.14% DTD, 2.27% weight), Ciena (+10.08%, 1.11% weight), and Tower Semiconductor (+8.78%, 1.32% weight), driven by strong demand for connectivity and specialty semiconductor solutions. Broadcom (+4.70%, 3.13% weight) and Eaton (+4.38%, 1.35% weight) also provided meaningful support.

• Key drags on AGIX DTD were Snowflake (-6.79%, 2.49% weight), ServiceNow (-6.04%, 1.88% weight), and Tempus AI (-6.02%, 0.90% weight), reflecting profit-taking in high-multiple software and AI application names. Alphabet (-3.86%, 3.45% weight) and Microsoft (-4.17%, 3.65% weight) also weighed on the fund.

News Highlights

1. Anthropic confidentially files for IPO: The AI leader's listing could unlock significant institutional capital for the AI sector and validate large-scale AI infrastructure spending. AGIX holdings such as Alphabet, Microsoft, and Broadcom—key Anthropic partners—could benefit from increased ecosystem activity.

2. Alphabet plans $80B equity raise for AI infrastructure: The massive capital raise underscores the insatiable demand for AI compute and data centers. As a top 3 AGIX holding (3.45% weight), Alphabet's aggressive buildout signals long-term growth but near-term dilution and margin pressure may weigh on the stock.

3. HPE earnings beat driven by AI servers: HPE's AI server revenue surged, sending shares up over 25% after-hours. While HPE is not an AGIX holding, the strong demand for on-premises AI infrastructure supports the thesis for AGIX hardware names like Broadcom, Arista, and Ciena.

4. Florida AG sues OpenAI over safety claims: The first state-led lawsuit against OpenAI introduces regulatory risk for the AI sector. While OpenAI is not publicly traded, increased scrutiny could slow adoption and impact AGIX infrastructure holdings like Microsoft and Alphabet that embed AI into their platforms.

5. Intel launches Xeon 6+ and new data center GPUs: Intel's 18A process and competitive power-efficiency specs challenge AMD and Nvidia in the data center. AGIX holdings like AMD and Nvidia face increased competition, though Intel's resurgence could benefit the broader hardware ecosystem.

RSS Highlights

1. Microsoft released two new small LLMs: MAI-Thinking-1 (3.5B params) and MAI-Code-1-Flash (500M params), trained on enterprise-grade data. This signals a trend toward smaller, specialized models that could lower inference costs and broaden AI adoption.

2. Meta reportedly plans multiple smart glasses launches this year, including 'Modelo', 'Luna', and 'Mojito VIP'. Continued investment in wearables highlights the growing importance of edge AI and on-device processing, benefiting AGIX hardware holdings like Qualcomm and Broadcom.

3. A critical analysis of AI ROI argues that opaque pricing and token-based billing make it hard for users to assess value. This could slow enterprise adoption if cost transparency does not improve, posing a risk to AGIX's infrastructure and application holdings.

4. Gary Marcus argues that AI commoditization will lead to price wars and thin profits, as no company has a unique data or technology moat. This view challenges the long-term margin assumptions baked into many high-multiple AI stocks in AGIX.

Performance & Benchmark

Attribution

Contribution of the three sectors within AGIX holdings today

Application: -0.46%

Hardware: 1.4%

Infrastructure: -0.83%

Top-performing stocks in AGIX holdings today

Astera Labs, Inc. (11.14%)
Ciena Corporation (10.08%)
TOWER SEMICONDUCTOR LTD (8.78%)
Flex Ltd. (7.72%)
Teradyne, Inc. (6.27%)
GlobalFoundries Inc. (5.84%)
Texas Instruments Incorporated (5.09%)
SANMINA CORP (5.09%)
ASML Holding N.V. (4.86%)
Broadcom Inc. (4.7%)
EATON CORP PLC (4.38%)
Vertiv Holdings Co (3.43%)
Samsung Electronics Co., Ltd. (3.3%)
TeraWulf Inc. (3.23%)
Apple Inc. (2.9%)
Micron Technology, Inc. (2.76%)
Arista Networks Inc (2.72%)
NEXTERA ENERGY INC (2.41%)
Advanced Micro Devices, Inc. (2.24%)

Worst-performing stocks in AGIX holdings today

Netflix, Inc. (-2.94%)
SAP SE (-3.0%)
Datadog, Inc. (-3.01%)
Samsara Inc. (-3.09%)
Riot Platforms, Inc. (-3.29%)
Alphabet Inc. (-3.86%)
CIRCLE INTERNET GROUP INC (-3.92%)
Microsoft Corporation (-4.17%)
Salesforce, Inc. (-4.18%)
Roblox Corporation (-4.26%)
CoreWeave, Inc. (-4.45%)
Palantir Technologies Inc. (-5.28%)
Shopify Inc. (-5.73%)
Tempus AI, Inc. (-6.02%)
ServiceNow, Inc. (-6.04%)
Snowflake Inc. (-6.79%)
Zscaler, Inc. (-7.42%)
NITTO BOSEKI CO LTD (-9.65%)

Top five listed stocks contributing the most within AGIX holdings today

Snowflake Inc. (-6.79%)
Microsoft Corporation (-4.17%)
Alphabet Inc. (-3.86%)
Palantir Technologies Inc. (-5.28%)
ServiceNow, Inc. (-6.04%)

Observation

The strongest movers across peer ETF holdings today are listed below:

News

Funding & M&A

Anthropic confidentially files for IPO amid rapid growth

AI company Anthropic has confidentially submitted its IPO application, with an expected market value exceeding $96.5 billion after listing. The company's latest financial report shows annualized revenue has exceeded $47 billion, and it may have achieved its first profitable quarter, with the Claude Code product generating over $2.5 billion in annualized revenue. Anthropic plans to launch a Mythos-level LLM model and will collaborate with AWS, Broadcom, Google, and SpaceX to build large-scale infrastructure. SpaceX also submitted its IPO application the same week, and OpenAI is expected to go public within the year. This listing will unlock opportunities for institutional investors while providing infrastructure financing support for Anthropic.

Rocket engine startup Impulse raises $500 million to hire people, not AI

Space mobility startup Impulse Space announced this week that it has completed a $500 million Series D funding round, led by 137 Ventures and BANNER VC, with participation from Founders Fund, Lux Capital, and Linse Capital. The funds will be used to hire between 4 and 200 new employees and accelerate the development and testing of high-maneuverability spacecraft such as Mira and Helios. The company's Mira platform is primarily aimed at purchases by the U.S. Space Force, while Helios is used to rapidly deliver satellites to high orbits. President Eric Romo, a SpaceX alumnus, stated that AI applications in integrated circuits and hardware engineering are still immature, and real physical testing remains irreplaceable. This funding round reflects investors' strong interest in aerospace and defense technology, and SpaceX's upcoming IPO is also injecting new momentum into the sector.

Technical Breakthroughs

Intel introduces Xeon 6+ server processors, previews upcoming graphics cards

Intel recently launched the Xeon 6+ series processors based on the latest Intel 18A process. The series is the first to adopt the Darkmont core design, focusing on power efficiency optimization. Among them, the Xeon 6990E+ features up to 288 cores, with significantly increased L3 cache capacity, and offers a 30% improvement in power efficiency and single-thread performance over the AMD Epyc 9965. Also released is the Crescent Island data center GPU, which uses the Xe 3P architecture and LPDDR5x memory, with a power efficiency improvement of up to 90% and support for multiple floating-point formats including MXFP4. The new generation Ethernet E835 network card delivers 1.9 times the performance of Nvidia's comparable products. This release demonstrates Intel's technological repositioning in the AI data center field.

MazeBolt launches RADAR VectorAI to test enterprise defenses against AI-generated DDoS attacks

MazeBolt Technologies Ltd. has launched the RADAR VectorAI module, which uses artificial intelligence to generate unprecedented DDoS attack vectors and validates them in real customer production environments to uncover configuration vulnerabilities in defense strategies. Positioned as the 'Mythos of the DDoS field,' the module aims to address the gap in network-layer protection validation beyond software vulnerability discovery, as seen with the Anthropic Claude Mythos model. VectorAI distinguishes between AI-orchestrated attacks and AI-generated attacks, runs on MazeBolt's existing RADAR platform, and continuously simulates attacks without disrupting services. Founder Matthew Andriani stated that AI-driven cyberattack capabilities are accelerating, and enterprises need to conduct targeted validation in their own environments. Recent attacks recorded by Cloudflare—2.01 billion requests per second—and Microsoft's Layer 7 attack analysis highlight the escalation of DDoS threats. MazeBolt will showcase the technology at Infosecurity Europe in London in June.

Product Launches

HPE posts huge earnings beat thanks to AI servers, sending its stock skyward

Hewlett Packard Enterprise (HPE) significantly exceeded expectations in its second fiscal quarter results, with AI server business experiencing explosive growth, driving revenue to $10.68 billion, up 40% year-over-year, and net profit turning around to a profit of $624 million. The Cloud & AI division generated $7.71 billion in revenue, with the server sub-segment exceeding expectations by $545 million. Benefiting from the Juniper consolidation, network business revenue surged 148% to $2.7 billion. The company raised its full-year earnings per share guidance to $3.35-3.45 and launched ProLiant servers based on Nvidia Vera CPUs, focusing on edge computing and agentic AI inference. CEO Antonio Neri emphasized strong demand from enterprise and government customers for on-premises AI infrastructure, with traditional server bookings reaching record highs. HPE's stock surged over 25% in after-hours trading, with year-to-date gains exceeding 95%. Despite memory cost pressures, the company continues to lead the AI infrastructure market with its high-margin AI product portfolio and execution.

Policy & Regulation

Florida Attorney General sues OpenAI and Sam Altman over claims that the technology is dangerous and exploits its users

On the same day, the U.S. state of Florida filed the first state-led lawsuit in the nation, accusing OpenAI and its CEO Sam Altman of ignoring safety warnings and causing harm to users through ChatGPT. The lawsuit claims the product contributed to a fatal shooting, led to suicides, and caused teenagers to become addicted to artificial emotional simulations. Florida Attorney General Ashley Moody stated in an 83-page complaint that OpenAI exploited user data and overlooked risks to increase market value. The lawsuit also seeks to hold Altman personally liable. This case comes as multiple states begin to strengthen oversight of AI companies, while the Trump administration adopts a more friendly stance. The move reflects the tension between national and state levels regarding AI risk prevention and innovation development.

融资收购

Alphabet unveils plan to sell $80 billion in shares to fund ongoing AI infrastructure buildout

Alphabet Inc. plans to raise $80 billion in equity capital through a stock offering to support its AI infrastructure construction. Berkshire Hathaway has agreed to invest $10 billion through a private placement. Alphabet stated that the funds will be used for data centers and world-class AI computing power investments to address unprecedented demand for AI services from enterprises and consumers. Alphabet is accelerating its efforts to catch up with competitors such as Amazon Web Services, Microsoft, and Meta, with the four major cloud giants expected to spend a combined over $700 billion on AI capital expenditures this year. Alphabet has raised its full-year capital expenditure forecast to $180-190 billion. CEO Sundar Pichai said computing capacity is his biggest concern. Alphabet has recently raised over $66 billion through multiple bond offerings, and will also issue $30 billion in underwritten stocks, $15 billion in mandatory convertible preferred shares, and sell $40 billion in Class A and Class C stocks in the third quarter through an ATM program. Goldman Sachs, JPMorgan Chase, and Morgan Stanley are acting as underwriters.

Alphabet plans to raise $80 billion to pay for AI buildout

Google parent company Alphabet announced on Monday that it will raise $8 billion through a stock offering to support large-scale AI infrastructure construction. The plan includes selling $10 billion in stock to Berkshire Hathaway. The company stated that current enterprise and consumer demand for AI services has exceeded its supply capacity, and it is rapidly expanding infrastructure to meet market growth. This move will maintain a healthy balance sheet rather than relying entirely on debt financing. Similar to other tech giants, Alphabet plans to invest $18-19 billion in computer capital expenditures in 2026, with global tech companies' AI-related capital expenditures expected to reach $700 billion. This financing will help strengthen its competitive advantage in the AI field.

RSS Subscription

An Ode to the Exacting Pedantry of Computers

The rigor of computer programming requires developers to explicitly specify variable types and operations when writing code, and this precision helps improve the quality of thinking and output. However, with the emergence of large language models (LLMs), this strict requirement for language and thinking is gradually disappearing, allowing users to realize ideas more quickly without having to think deeply about details. While this change improves efficiency, it may also lead to unclear intentions, thereby affecting user experience.

Why things will eventually fall apart

As competition in the artificial intelligence field intensifies, major companies like Google hold market share, but due to a lack of unique technology and data moats, the market will struggle to form monopolies, ultimately leading to price wars and thin profits. More and more investors are paying attention to this phenomenon, and many experts doubt the commercial viability of the large language model industry, believing that the current economic model is unsustainable and that retail investors and pension funds may be affected.

Logic for Programmers extra credits

This week, because I was preparing for a conference in Budapest, I did not publish a formal newsletter, but I prepared some supplementary materials covering several interesting topics, including sorting computations for concurrent processes, quantification of 'function sets' in first-order logic, Barbara Liskov's 'history rule' for subtypes, and total and partial orders on sets, totaling about 2000-3000 words. The supplementary materials have not been strictly edited and may contain some errors. See you next week!

AI Doesn't Have ROI

Recently, the AI industry has faced a lack of cost transparency, with users often unable to accurately assess their expenses and return on investment when using AI subscription services. For example, after Microsoft's GitHub Copilot switched to token-based billing, many users consumed a large number of tokens in a short time, leading to customer anger. AI companies deliberately obscure fee structures, making it difficult for users to understand actual costs, a phenomenon that is widespread across the industry.

Using FourSquare's API to post location checkins to social media

Developers can obtain the SwarmToSocial code via GitLab, with 10,000 free API calls per month, suitable for personal use. FourSquare's CheckIn documentation is easy to understand; developers need to create an OAuth app to obtain a client ID and secret, and then can publish new check-ins through the API.

Cyrix 486DLC CPU: Introduced June 1992

In June 1992, Cyrix launched the 486DLC CPU, a processor combining features of the 386DX and 486SX, with an initial price of $119 and a maximum speed of 40 MHz. Although the 486DLC performed close to a 25 MHz 486SX, its smaller 1K L1 cache and slower 386 bus made it inferior to Intel's 486 series overall. The release of the 486DLC allowed motherboard manufacturers to convert 386 motherboards into 486 motherboards via a BIOS upgrade, extending product lifecycles.

Pluralistic: The tedious power of storytelling (02 Jun 2026) must-we-pretend

In a lecture on creativity and art, Brian Eno proposed that art is "everything you don't have to do," emphasizing that the purpose of art is to make others feel something. He pointed out that AI-generated works lack true intent, diluting the emotions they express and ultimately making it difficult for audiences to resonate with them.

Is the Monaco Grand Prix decided at qualifying?

In the Monaco Grand Prix, the impact of starting position on race results is not as significant as some drivers claim. Over the past 25 years, only 80% of races have been won by drivers starting from the front row, far below the claimed 90%. Additionally, the overall capability of the driver and car (i.e., "equipment ability") plays a more important role, with about 70% of the points difference between drivers being related to their driving speed.

The placeholder name for the Windows 8 experience was “modern”

During the development of Windows 8, the team named the new system "Modern Experience" to distinguish it from Windows 7's "Classic Experience." To simplify naming, many components were prefixed with "Mo," such as "Modern Shell" shortened to "MoSh," "Start Menu" called "MoGo," and the new browser referred to as "MoBro." Windows 8 was officially released on June 1, 2011, fifteen years ago.

Microsoft's new MAI models

On June 2, 2026, Microsoft released two new text large language models: MAI-Thinking-1 (3.5 billion parameters, aimed at "select early partners") and MAI-Code-1-Flash (500 million parameters, designed for GitHub Copilot and VS Code to deliver high performance and low cost). MAI-Thinking-1 outperformed Sonnet 4.6 in blind tests, and both models were trained on enterprise-grade, clean, and legally sourced data.

Meta Reportedly Has a Slew of New Smart Glasses Planned for This Year

Meta maintains its lead in smart glasses and plans to launch multiple new products. According to The Information, Meta is expected to release a smart glasses model called "Modelo" next month, and also plans to launch two new models in the fall, named "Luna" and "RBM2 Refresh." By December, Meta will also release a pair of glasses called "Mojito VIP," and is testing two prototypes, "Artemis" and "SSG," with the latter featuring an always-on camera for "hyper-sensing" capabilities aimed at providing continuous visual information. Despite challenges such as privacy and battery life, Meta remains committed to advancing smart glasses development.

Scott Pelley Accuses CBS News Boss of ‘Murdering’ ‘60 Minutes’

CBS News experienced new turmoil on "60 Minutes," with host Scott Pelley sharply criticizing new executive producer Nick Bilton during a staff meeting and accusing network editor-in-chief Barry Weiss of "murdering" the long-running program. Although "60 Minutes" won two News Emmy Awards on the eve of the layoffs and saw a 9% year-over-year increase in ratings, Pelley still expressed concerns about the show's future, believing that both Weiss and Bilton lack sufficient qualifications.

Apple, the Anti-‘Metaverse’ VR Company

Meta is considering layoffs within its Reality Labs division's metaverse team, with a reduction of 10% to 30% expected, possibly implemented next month. Despite this, Meta does not plan to abandon metaverse development, but instead intends to redirect the saved funds toward augmented reality glasses development.

‘The Metaverse Fever Dream’

The concept of Meta's "metaverse" was first introduced by Neal Stephenson in his 1992 novel "Snow Crash," and was later preliminarily demonstrated in the 2003 game "Second Life." In 2012, Oculus was founded and released a developer version of its virtual reality headset in 2013, before being acquired by Facebook. Although Facebook stated in 2018 that it viewed virtual reality as a new paradigm for future computing, enthusiasm for the metaverse has waned, leading many to question the practical significance and feasibility of the concept.

The Metaverse Was Snake Oil for Isolation

Nick Heer published a thoughtful article on Pixel Envy about the rise and fall of the "metaverse" hype, noting that since Matthew Ball predicted in 2020 that the metaverse would create trillions of dollars in value, Silicon Valley's obsession with it has deepened. However, as the pandemic unfolded, many found themselves living in an increasingly virtual space, and the gap between the ideal of the metaverse and reality gradually became apparent, causing many once-enthusiastic futurists to distance themselves from the concept.

California Brown Pelican

Today I attended the Microsoft Build conference at Fort Mason in San Francisco, where I could see California brown pelicans diving into the water.

Three Ways to Get Paid

Jason Zweig shared three survival rules his father taught him in 2018: 1) Lying to those who want to be deceived can make you rich; 2) Telling the truth to those who want to hear it can earn you a living; 3) Telling the truth to those who want to be deceived will bankrupt you. These ideas echo Om Malik's discussion about 'how liars and salesmen succeed.'

The First-Time-Buyer-Discount Dickover Scheme

Neil Panchal criticized a promotional strategy targeting first-time buyers on Twitter, calling it a 'dickover' because it attracts new customers with a 10% discount while making existing customers feel neglected. Although this tactic may draw some users in the short term, it can also create a negative impression of the brand and reduce their future purchase intent. The author shared his experience shopping on the New Era website, saying he would rather pay a higher price than deal with the hassle of such marketing tactics.

Pasted File Editor

Pasted File Editor is a tool that allows you to paste large amounts of text into claude.ai, which will recognize it as a large paste and convert it into a file attachment. Codex Desktop is developing a similar prototype that supports directly opening files (including images displayed as thumbnails) or dragging and dropping files into the text area.

[Sponsor] Mux — Video for Developers

Mux provides video processing solutions to help teams efficiently handle video streaming, analytics, and transformation, enabling fast builds. Users can leverage the Mux platform for video creation, real-time analytics, and automatic caption generation, and can start without a credit card. Through successful case studies with multiple companies, Mux demonstrates significant results in improving user engagement and sales conversion rates.

When su replaced login for becoming another Unix login

Due to attacks from high-traffic crawlers, some browser versions (especially older Chrome) are considered suspicious, making it impossible to access my blog or related wikis. If you are using a current version of your browser, please contact me and provide your browser information. It is recommended to use archive.org for page archiving, as other archiving services may misuse old user agents, leading to access errors.

‘If You Take the Weasel Job Then You Must Be the Weasel’

Nick Bilton has been appointed as the new head of 60 Minutes, raising concerns about CBS News' credibility. Due to his lack of television news experience and background, many journalists are uneasy that he will continue to weaken editorial independence, especially under the influence of Bari Weiss, as the program shifts toward catering to power rather than truthful reporting. True investigative journalism requires a determination to confront misinformation, not an attitude that panders to corporate interests.