AI News Report — 2026-06-29 (Mon)

AI industry trend report | June 29, 2026

Overview

The biggest topic of conversation is the geopolitical movement in which Anthropic's Claude Fable 5 has been revived from a two-week export ban by the US government, and Europe is moving to attract Anthropic. xAI announced Grok 4.5 as a V9 model with 1.5 trillion parameters and started internal testing at Tesla and SpaceX. On the other hand, the lack of AI supply is emerging as a cross-industry issue, with Google restricting Meta's use of Gemini.

Anthropic / Claude trends

The biggest focus is on lifting export restrictions on Claude Fable 5. There have been a number of reports that the US government has approved the revival of Fable 5 after a security review. During the ban period, Austria was lobbying the EU to host Anthropic in Europe, making it clear that US AI export restrictions are having the opposite effect of promoting European independence. On the technical side, headlines read, ``Claude Opus 4.8 solves AI's biggest problems,'' and there are reports that Claude AI has achieved 20 times the performance of the robotics team. The Bulletin of the Atomic Scientists discusses why high-performance models like Claude Fable and Mythos don't fit within traditional export control frameworks, and the legal vacuum in AI regulation has sparked debate.

OpenAI trends

Focus on hardware strategy and capital policy. OpenAI and Broadcom's joint announcement of the first custom AI chip to compete with NVIDIA sent ripples through the semiconductor industry. There are also reports of negotiations with Qualcomm regarding smartphone AI chips, and OpenAI's desire for vertical hardware integration is accelerating. Regarding the IPO, an article appeared stating that SpaceX is considering postponing its IPO due to stock price fluctuations. In terms of human resources, it was reported that Apple Vision Pro's VP position was transferred to OpenAI, and OpenAI continues to have an advantage in the competition to acquire talented personnel. Sam Altman continues his global expansion by speaking at the AI Impact Summit in India.

Google AI trends

Tight supply and demand symbolized by supply constraints and Gemini access restrictions. The Financial Times reported that Google had restricted Meta's Gemini API access, and multiple media outlets followed suit and confirmed this. The reason is a lack of capacity, which clearly shows that the demand for AI exceeds the supply, and even Google can't keep up. It was revealed that six researchers from DeepMind's Coding department were transferred to Meta, OpenAI, and Anthropic, and the impact on Alphabet stock is being discussed. There have also been articles criticizing Demis Hassabis' management decisions, and DeepMind's talent loss problem is serious.

Trends of other AI companies

xAI: Announced Grok 4.5 as a V9 model with 1.5 trillion parameters. Elon Musk himself announced the start of private betas with Tesla and SpaceX. It has also been revealed that the FTC has approved Musk's acquisition of AI networking startup "Mesh". Meanwhile, Yann LeCun's public statement that xAI is a failure and Musk is having a hard time hiring top talent has caused a stir within the industry.

DeepSeek: Announced that it released "DSpark" that utilizes Speculative Decoding and increased the inference speed of V4 models by up to 85%. Chinese media has positioned this as the "next DeepSeek moment," suggesting a response to the U.S. technology blockade.

Microsoft: Enhances Copilot's capabilities for Excel, adding reusable skills and data connectors. On the other hand, Edge's AI history search function has been abolished. There are also warning reports that prompt injection is emerging as an attack targeting enterprise AI RAG pipeline agents.

Meta: Appeared as a person who received gemini access restrictions from Google. Employee tracking program suspended after internal data breach. There have been reports of ``snacks being provided'' as a measure against declining morale, and articles that suggest management turmoil are prominent. There is also a critical report (WION) stating that ``Meta's AI has destroyed the lives of 2 million people.''

Remarks and trends of AI celebrities

Sam Altman (OpenAI CEO): He spoke at the AI Impact Summit in India and demonstrated his stance on global AI deployment. GIGAZINE features a feud between Altman and Dario Amodei that has spanned over 10 years.

Elon Musk (xAI / Tesla CEO): Announced the start of private testing of Grok 4.5 in the form of a personal leak. There are also reports that SpaceX has invested $60 billion in AI, and the question ``Are we building the next Amazon?'' has become a hot topic in the market. SoftBank CEO Masayoshi Son questioned Musk's vision, asking, ``What is the significance of an AI data center in space?''

Jensen Huang (NVIDIA CEO): ``No company can match NVIDIA's cost performance.'' We remain bullish even though NVIDIA stock is down 18% in 2026. He expressed the view that the market environment where demand for AI exceeds infrastructure supply will work in NVIDIA's favor.

Dario Amodei (Anthropic CEO): It attracted the dual attention of Fable 5's unbanning and EU invitation. Technical results regarding Claude AI's robotics performance were also reported.

Yann LeCun (Meta Chief AI Scientist): Declares xAI a "failure" and points out Musk's recruitment difficulties. He also mentioned the shift of top developers to physical AI and appealed for a shift from AI development centered on chatbots.

Sundar Pichai (Google/Alphabet CEO): Faced with two challenges: Gemini's supply and demand problem and DeepMind's talent outflow. An article in which a former Google employee talks about the "crazy traits" that Larry Page, Sergey Brin, and Sundar Pichai have in common attracted attention.

Mark Zuckerberg (Meta CEO): Social criticism of Meta's AI services is increasing, with critical articles calling for ``Zuckerberg to resign'' appearing.

Demis Hassabis (Google DeepMind CEO): The decision to pivot to the coding department has been criticized for causing an exodus of researchers, and an essay has even appeared calling for demotion or firing.

Breakthroughs and innovations in AI technology

Grok 4.5 (xAI): The 1.5 trillion parameter V9 model is larger than the GPT-4o model and indicates the direction of larger scale MoE architecture. DeepSeek DSpark: As a practical implementation of Speculative Decoding, it is noteworthy that it has achieved a significant reduction in inference cost (up to 85% speedup). In the edge AI field, Qualcomm is making a full-fledged move into the AI data center field to compete with NVIDIA, and along with OpenAI and Broadcom's joint development of custom chips, the multipolarization of the semiconductor ecosystem is accelerating. Reports that Apple iOS 27 will allow users to select third-party AI models signal an important shift in the openness of mobile AI.

AI Regulation/Policy/Social Impact

The most important topic is Claude Fable 5 export regulations and European movements. The precedent set by the U.S. government by temporarily banning the export of high-performance AI models for security reasons and then reinstating them will have a major impact on the framework of future AI export regulations. The fact that the EU, led by Austria, is considering bringing Anthropic to Europe is a sign of geopolitical competition over "digital sovereignty" in AI. Pope Leo XIV's declaration that ``AI cannot be morally neutral: Ethics is embedded in all algorithms'' is another notable religious and philosophical position statement. The issue of AI copyrights in entertainment companies remains unresolved and is expanding, and there is also an example of the FBI using AI to investigate the attack on the White Palace Banquet Party.

Integrated analysis: What is happening now

Supply constraints, geopolitics, and hardware independence are converging as three major themes. The situation in which Google is forced to restrict Meta's use of Gemini shows the reality that cloud AI supply is not keeping up with demand. This is completely consistent with Jensen Huang's assertion that demand is outstripping supply.

Anthropic's Fable 5 ban/unban episode exposed weaknesses in US AI governance. Europe's decision to take advantage of this opportunity to host a bid could accelerate competition between the United States and Europe regarding AI governance.

xAI's Grok 4.5 announcement and DeepSeek's DSpark show that players other than OpenAI/Anthropic are seriously catching up in both high-performance models and efficient inference. In particular, DeepSeek's inference cost reduction approach serves as an alternative to costly proprietary models.

Movement to break away from dependence on NVIDIA is also clear. OpenAI+Broadcom's custom chips, Qualcomm's entry into data centers, and Apple's rising device prices (due to a shortage of AI chips) show that the reorganization of the semiconductor ecosystem is progressing as a real demand.

Human resource mobility is also an important signal. The movement of human resources between companies, such as the exodus of researchers from DeepMind to Meta, OpenAI, and Anthropic, and Apple executives changing jobs to OpenAI, is causing a reallocation of technological capabilities.

Future points of interest

International expansion of Claude Fable 5 and establishment of a permanent framework for export controls — Whether the move to attract Europe comes to fruition or the US institutionalizes export controls may be decided in the coming weeks.

Assessing the ability of OpenAI + Broadcom custom chips and Qualcomm's strategy to compete with NVIDIA — Whether or not it can compete with NVIDIA's cost performance at the actual product level could fundamentally change the industry's dependence on semiconductors.

The impact of Google DeepMind's talent outflow and Gemini supply constraints — How Alphabet resolves the double whammy of internal organizational collapse and external supply shortages will determine the competitiveness of Google AI.