2026-06-02 AI Daily | Stargate Groundbreaking and a Year of Scientific Breakthroughs: AI Infrastructure and Platform Ecosystems Accelerate in Sync Link to heading

OpenAI breaks ground on the 1GW Stargate data center in Michigan, which, along with NVIDIA’s Vera Rubin architecture, propels the computing foundation’s evolution toward both “extreme scale” and “extreme edge.” At the Snowflake Summit, Sam Altman defined 2026 as the year of AI-driven scientific breakthroughs, as enterprise-grade Agentic AI moves toward official delivery. Meanwhile, Apple’s launch of the genai.apple.com domain hints at the release of a full-stack generative AI, and Microsoft’s collaboration with NVIDIA to reinvent Windows as an Agentic OS brings the battle for the agent platform entry point to a one-week countdown.

📖 Deep Dive: This Issue’s Watch List Link to heading

Today’s information flow contains three interconnected deep-dive themes worth noting for both engineering and research teams.

The first is the comprehensive restructuring of the AI computing foundation. NVIDIA just unveiled its Vera Rubin architecture and RTX Spark superchip at Computex, followed by Jeff Dean’s in-depth explanation of how a million-fold leap in computing power will rewrite research paradigms. Simultaneously, OpenAI began construction on its 1GW Stargate data center in Michigan. These events combined indicate that computing power supply, from cloud to edge, is advancing in two directions simultaneously: toward the “extremely large” and the “extremely close.”

The second is the “fulfillment of expectations” for AI-driven scientific discovery. At the Snowflake Summit, Sam Altman explicitly defined 2026 as the year of AI scientific breakthroughs, while Demis Hassabis discussed, from another perspective, how AI can solve fundamental problems in physics. This is not just a slogan but a substantive projection based on the trend of officially delivering Agentic AI integrated with enterprise data.

The third is the shadow war over the terminal agent ecosystem. Apple’s sudden launch of the genai.apple.com domain hints that WWDC will unveil a full-stack GenAI strategy, while Microsoft and NVIDIA are again joining forces to reinvent Windows as an Agentic AI OS. The race for platform-level agent entry points has entered a one-week countdown.

🌐 AI Hot Topics on X Link to heading

Topic 1: Perplexity Unveils Search as Code for Smarter AI Agents Link to heading

• Category: AI · News
• Overview: Trending Time:, Related Posts: 127
• What it is: Perplexity released “Search as Code,” a feature that allows AI agents to break down search tasks into executable, composable code blocks to enhance their search and reasoning efficiency.
• Why it matters: This move transforms search from a black-box Q&A process into programmable, reusable infrastructure. It enables AI agents to dynamically build and optimize search pipelines like calling an API, significantly improving the accuracy and controllability of complex information retrieval, multi-step reasoning, and tool use.
• Discussion Overview: The discussion primarily revolves around three points: first, whether “Search as Code” can truly replace traditional RAG (Retrieval-Augmented Generation) workflows and improve agent reliability in vertical domains; second, developers are skeptical about the pricing and ecosystem impact of Perplexity packaging search capabilities as a paid API; third, some users worry this will further blur the lines between search engines and AI agents, leading to disputes over content copyright and traffic attribution.

Topic 2: OpenAI’s GPT-5.5 and Codex Launch on Amazon Bedrock for All AWS Users Link to heading

• Category: AI · News
• Overview: Trending Time:, Related Posts: 736
• What it is: OpenAI has officially made its latest GPT-5.5 and Codex models available to all AWS users through Amazon Bedrock.
• Why it matters: This marks the first time OpenAI’s advanced models are deeply embedded in a major cloud service platform, significantly lowering the barrier for enterprises to use them and potentially accelerating the adoption and scaling of AI applications in enterprise scenarios.
• Discussion Overview: Discussions on X focus on the actual performance improvements of GPT-5.5 over its predecessors, the code generation quality of Codex, its competitive relationship with AWS’s own models and Anthropic’s Claude, and the pricing strategy and usage limits for the new models on Bedrock.

Topic 3: Anthropic Shares Advanced Claude AI Tips from Developer Conference Videos Link to heading

• Category: AI · News
• Overview: Trending Time: 6 hours ago, Related Posts: 763
• What it is: Anthropic shared several advanced Claude AI usage tips in videos from its developer conference.
• Why it matters: This indicates that leading AI companies are shifting from merely releasing models to deeply empowering developers, which helps lower the barrier to implementing complex AI and promotes the widespread adoption of agent-based applications.
• Discussion Overview: The main topics of discussion on X include: the degree to which these tips improve actual development efficiency, the varying utility of different video content, and whether this kind of public sharing can truly bridge the gap between model capabilities and user applications.

Summary of AI Public Opinion on X Today Link to heading

Today’s main narrative clearly points to the AI industry’s shift from a competition over model capabilities to a systematic battle for infrastructure and developer ecosystems. The consensus is that whether it’s Perplexity refactoring search into programmable code blocks, OpenAI deeply embedding its top-tier models into the AWS cloud platform, or Anthropic publishing advanced usage techniques, the giants are all striving to standardize and toolify complex AI capabilities to lower the barrier for enterprise and developer adoption. The disagreements lie in different judgments about the implementation paths and ecological impact. For example, developers are skeptical about whether Perplexity’s “search-as-code” can truly replace traditional RAG workflows and are wary of its API pricing model. At the same time, there is heated discussion about the competitive relationship between OpenAI on AWS and its own models, as well as Claude. A prominent potential risk is that as the boundaries between search, programming, and agents become increasingly blurred—especially after search capabilities are made programmable and commercially packaged—it could intensify content copyright disputes and structural conflicts over internet traffic attribution. This will be a challenge that the era of reusable infrastructure must face.

💡 Influencer Insights Link to heading

AI Industry Daily: June 1, 2026 Trend Insights Link to heading

I. Today’s Key Tech Trends and Product Hotspots Link to heading

1. Explosion of Agent Product Forms: From Tools to “Operating Systems” Link to heading

• The intense competition between Codex and Claude Code has become the absolute focus. @dotey observed a large number of users “jumping ship” to Codex, praising its excellent Chrome integration and parallel background task capabilities; while @Pluvio9yte, after testing Claude Opus 4.8, believes the “improvement is not obvious, and considering the overall price, would still choose GPT-5.5.”
• Cursor doubles down on the enterprise market: Announced increased user quotas for Teams and introduced Premium seats (5x usage for only 3x the cost). @dotey commented that its multitask mode and Plan mode have good stability, but the lack of a /goal command and Chrome use capability is a shortcoming.
• Coze 3.0’s breakthrough in China: @Pluvio9yte calls it a “clear leader in China,” with core highlights being built-in Claude Code and Codex, supporting cloud-based Agents calling local Codex, achieving the ability to “assemble an Agent army in ten minutes.”

2. Practical Application of On-device LLMs Accelerates Link to heading

• Mac becomes the main platform for on-device development: @zhixianio demonstrated a Mac running 3 mainstream on-device models simultaneously, even using a model to “thaw a rice ball” (a vision + reasoning scenario), calling the fan noise “melodious.”
• MiniCPM5-1B tops the small model leaderboard: @OpenBMB released the most powerful open-source base model under 2B, with an AA index of 17.9, surpassing Qwen3.5-2B.
• AMD Ryzen AI Halo enters the scene: @zhixianio forwarded AMD’s mini PC solution, pre-installed with ROCm and optimized workflows, “another platform for on-device models has arrived.”

3. Deep Anxiety Over AI Programming Costs and Business Models Link to heading

• @ruanyf drops a bombshell observation: The founder of OpenClaw consumes 603 billion Tokens per month (valued at $1.3 million), sparking a sharp discussion on “AI programming being more expensive than human programmers”—“even with cheaper models, it would still cost 2-3 million RMB a year.”
• Enterprise consulting becomes the new battlefield: @Pluvio9yte pointed out the collaboration between OpenAI DeployCo ($4 billion) and Anthropic×KPMG, marking a shift for model companies from “selling APIs” to “sending people into enterprises to deconstruct workflows and integrate with legacy systems.”

II. Noteworthy Unique Perspectives and Industry Foresight Link to heading

III. Recommended Tools & Resources Link to heading

Development Tools Link to heading

Skill/Automation Solutions Link to heading

Data & Content Resources Link to heading

IV. What to Watch for Tomorrow Link to heading

• OpenAI Livestream (June 2, 8:30 am PT): Preview of major updates for Codex and the OpenAI Platform (@derrickcchoi/@dotey)
• Differences in Claude’s Credit Reset Policy: Claude uses a “full credit for the remaining time” model instead of Codex’s “7-day rollover,” which @dotey described as “more generous.”

Report compiled from public posts on X from May 30 to June 1, 2026

📚 Appendix: Today’s Watch List Source Updates Link to heading

Time window: Last 1 day; Covers 16 sources; 8 updates in total

Stratechery by Ben Thompson (A_full) Link to heading

• YouTubers Win the Box Office, Goodbye Gatekeepers, The YouTube Bar
  • Publication Time: 2026-06-01 18:00 Beijing Time
  • Summary: Ben Thompson analyzes how YouTubers are dominating the box office, arguing that success on YouTube has a much higher bar than current Hollywood, signaling the end of traditional gatekeepers.
  • EN Highlights:
• Success on YouTube now represents a higher bar than traditional Hollywood gatekeeping.

OpenAI Blog (A_full) Link to heading

• Building the infrastructure for the Intelligence Age in Michigan
  • Release Time: 2026-06-01 20:00 Beijing Time
  • Summary: OpenAI has launched a 1GW data center project in Michigan as part of the Stargate initiative. The project aims to build AI infrastructure to expand access, create jobs, and support the community. It has received strong support from the Governor of Michigan and promises that local residents will not have to bear the infrastructure costs.
  • EN Key Points:
    • OpenAI breaks ground on a 1GW data center project in Michigan as part of Stargate.

Two Minute Papers (B_intro+search) Link to heading

• What Happens After A 1,000,000x AI Compute Leap? | Jeff Dean

  • Release Time: 2026-06-01 23:41 Beijing Time
  • Summary: What will happen after a 1,000,000x leap in AI computation? In an interview, Jeff Dean explores the impact of the explosive growth in large model computing power on scientific research and practical applications.
  • EN Key Points:
    • Jeff Dean discusses the implications of a massive leap in AI compute capabilities.

• Einstein vs Feynman: AI’s Role in Future Scientific Breakthroughs

  • Release Time: 2026-06-01 08:16 Beijing Time
  • Summary: In an interview, Demis Hassabis discusses the role of AI in future scientific breakthroughs and how AI can help scientists solve complex problems, such as those in physics.
  • EN Key Points:
    • Demis Hassabis discusses how AI will drive future science.

OpenClaw Deep Research (A_full) Link to heading

• Apple WWDC 2026 Preview: genai.apple.com Domain Goes Live, Full-Stack GenAI Is Coming

  • Release Time: 2026-06-02 08:00 Beijing Time
  • Summary: Apple has been detected launching the genai.apple.com subdomain, hinting at a major AI announcement at next week’s WWDC. Bloomberg reports that Apple is building a full-stack generative AI ecosystem, with a focus on on-device processing and developer tools. Analysts predict Apple could become the first company to exceed a $6 trillion market cap, with AI as the core driver.
  • EN Key Points:
    • Apple launches genai.apple.com ahead of WWDC 2026.
    • Rumors of a full-stack generative AI ecosystem integration.
    • Focus on on-device privacy and specialized developer AI tools.

• Snowflake Summit 2026: Sam Altman and Daniela Amodei Discuss “AI Discovery Breakthroughs”

  • Release Time: 2026-06-02 01:00 Beijing Time
  • Summary: At the Snowflake Summit 2026 in San Francisco, OpenAI CEO Sam Altman predicted that 2026 will be a breakthrough year for AI-driven scientific discovery. The conference theme, “Making AI Real for Business,” signals a shift for enterprise data teams from discussion to the formal delivery of Agentic AI. Anthropic co-founder Daniela Amodei emphasized the central role of data in the next generation of AI transformation. OpenAI’s frontier models and Codex are now officially available in the AWS environment.
  • EN Key Points:

• Sam Altman predicts 2026 as a breakthrough year for AI-driven discovery.

• Snowflake Summit focuses on operationalizing Agentic AI for enterprise.

• Daniela Amodei (Anthropic) discusses the future of AI innovation.

• OpenAI frontier models and Codex now available on AWS.

• NVIDIA Computex 2026: Jensen Huang Unveils RTX Spark Superchip and Vera Rubin Architecture

  • Release Time: 2026-06-01 20:00 Beijing Time
  • Summary: Jensen Huang delivered a keynote at Computex 2026 in Taipei, unveiling the RTX Spark superchip platform and the next-generation Vera Rubin AI computing architecture. NVIDIA announced a deep collaboration with Microsoft aimed at reshaping Windows into an “Agentic AI OS.” Additionally, the NVIDIA OpenShell runtime, DLSS 4.5 Ray Reconstruction, and the revolutionary Vera CPU were introduced. Huang emphasized that 2026 is the “Year of Autonomous Agents,” and that the construction of computing power has entered the “AI factory” era.
  • Key Points:
    • NVIDIA unveils RTX Spark Superchip platform and Vera Rubin architecture.
    • Partnership with Microsoft to create an Agentic AI OS for Windows.
    • Introduction of NVIDIA OpenShell runtime and Vera CPU.
    • 2026 declared as the ‘Year of Autonomous Agents’.

X Platform (B_intro+search) Link to heading

• OpenClaw Releases Faster, Lighter AI Agent Update 2026.5.28
  • Release Time: 2026-06-01 16:00 Beijing Time
  • Summary: OpenClaw has released the 2026.5.28 update, delivering a faster and more lightweight AI Agent. The community has shown interest in its performance improvements and potential for on-device deployment.
  • Key Points:
    • OpenClaw releases a major performance update.
    • Focus on lightweight and fast execution for edge devices.