AI Capital Race Accelerates | Rewire News Morning Brief

Anthropic invests 200 billion in Google Cloud for five years, OpenAI spends 50 billion dollars annually on computing power, Apple pays 250 million dollars in settlement for AI shortcomings. Capital density crushes independent participants.

1｜Anthropic bets 200 billion on Google Cloud, opens Wall Street with AI agents

Anthropic signs a five-year, $200 billion cloud computing agreement with Google, becoming Google Cloud’s largest single customer commitment, accounting for over 40% of Google’s deferred revenue. Also signs a multi-gigawatt TPU capacity agreement, launching in 2027. Alphabet’s reverse investment in Anthropic could reach up to 40 billion dollars.

On the same day, Anthropic launches ten banking AI agents and Claude Opus 4.7 at its New York event, with JPMorgan CEO Jamie Dimon endorsing “Every dollar spent on AI is worth it.” Reuters reports that OpenAI and Anthropic are negotiating joint ventures to acquire multiple AI service companies, aiming to accelerate enterprise penetration. Infrastructure binding is used as an entry ticket into the industry; laboratory competition has shifted from model capabilities to distribution channels.

(Source: The Information / Reuters / Fortune / Axios)

2｜White House AI policy does a 180: after abolishing the executive order, rebuilds regulation under the guise of national security

Google, Microsoft, xAI, and the Department of Commerce’s NIST CAISI sign agreements to accept federal security assessments before releasing new AI models. Models are submitted for testing in a protected state to identify cyberattacks and military misuse risks. OpenAI and Anthropic renegotiate existing terms to align with Trump’s AI action plan. All five leading labs are now under pre-release review systems; CAISI has completed 40 assessments since 2024.

This marks a quiet stance reversal. After Trump took office, the Biden AI executive order was revoked, leading Silicon Valley to believe federal regulation had ended. Now, regulation returns under a new guise: the framework is “national security,” mechanisms are voluntary rather than mandatory. The substance remains largely unchanged, but discourse shifts from academic safety communities to defense and intelligence agencies.

(Source: NIST / Reuters / The Verge / Al Jazeera)

3｜Coinbase cuts 14% of staff, crypto industry becomes a testing ground for AI restructuring

Coinbase lays off about 700 employees, with CEO Armstrong attributing it to market downturn and AI efficiency gains. Management levels are compressed to five tiers below the CEO, with all managers also serving as individual contributors; some teams are consolidated into single roles combining engineering, design, and product. Severance pay ranges from $50 million to $60 million, and the stock price rose on the same day.

Coinbase is not alone. Bloomberg reports that the crypto industry is experiencing a wave of AI-driven layoffs. PayPal announces a similar transformation aiming for $1.5 billion annual savings. Polymarket traders bet on more tech layoffs to come. Single-person, multi-role teams are emerging—not just cyclical adjustments, but a permanent organizational shift.

(Source: CoinDesk / Reuters / Bloomberg / TechCrunch / CNBC)

4｜Apple abandons AI in-house development, iOS 27 fully opens to competitors

Bloomberg reports that in this fall’s iOS 27, Apple will allow users to select third-party AI models to power Apple Intelligence features. Google and Anthropic are already testing internally. Third-party models will be integrated via “Extensions” in system settings, capable of powering Siri, writing tools, and image generation. The exclusive ChatGPT position ends, with actual usage below Apple and OpenAI’s expectations.

On the same day, Apple settles a class-action lawsuit over AI feature misrepresentation for $250 million, with iPhone 16 and 15 Pro users eligible for up to $95 per device. Apple is also considering involving Intel and Samsung in U.S. chip manufacturing to diversify reliance on TSMC, with Intel’s stock jumping over 12% that day. The AI strategy shifts from in-house R&D to platformization, essentially acknowledging its models cannot stand alone.

(Source: Bloomberg / TechCrunch / NYT / Tom’s Hardware)

5｜Huawei’s AI chip revenue hits $12 billion, NVIDIA’s China market share drops to zero

Huawei expects AI processor revenue to reach approximately $12 billion by 2026, a 60% increase year-over-year. The growth is based on existing orders from Alibaba, ByteDance, Tencent; Ascend 950PR entered mass production last month. Morgan Stanley estimates China’s AI chip market will reach $67 billion by 2030, with Huawei potentially capturing 60% of the market by year-end.

On the other hand, Huang Renxun explicitly states that China should not have access to Blackwell or Rubin series GPUs, claiming the U.S. should maintain “first, most, and best” in AI hardware. Export controls have reduced NVIDIA’s China market share to zero; Huawei is filling the vacuum by building a self-sustaining ecosystem. NVIDIA, in turn, publicly states that “China should not have” certain hardware. Decoupling is now a fact; the question is where each ecosystem’s efficiency limits lie.

(Source: Tom’s Hardware / Digitimes / Morgan Stanley / WCCFTech)

Also worth knowing ↓

OpenAI spends 50 billion dollars annually on computing power. Co-founder Brockman testifies in Musk’s lawsuit, stating OpenAI’s 2026 computing expenses will reach $50 billion, up from $30 million in 2017. The company has raised over $100 billion in total funding and is preparing for a potential trillion-dollar valuation IPO. (Source: Bloomberg / Quartz)

OpenAI releases GPT-5.5 Instant, with hallucination rates down over 50%. This model replaces GPT-3.5 Instant as ChatGPT’s default. OpenAI claims hallucination rates in high-risk scenarios like medicine, law, and finance are reduced by 52.5%. (Source: TechCrunch / The Verge)

SEC proposes ending the quarterly reporting system for public companies. The mandatory quarterly reports, in place for 55 years, would become optional semi-annual reports. Trump had this idea during his first term. Investors worry about decreased market transparency; some companies welcome reduced short-term performance pressure. A 60-day public comment period has begun. (Source: SEC / WSJ / CNBC)

DeepMind employees in the UK vote to form a union protesting military AI contracts. This will be the first union at a leading AI lab. The trigger was Google’s agreement with the Pentagon to allow the U.S. Department of Defense to use Gemini models in classified military networks. (Source: Fortune / Wired)

a16z crypto raises $2.2 billion against the trend. This is the fifth fund for a16z’s crypto division, half the size of its record-breaking 2022 fourth fund. While many crypto VCs shift to AI investments, a16z crypto continues to bet on crypto. CTO Lazzarin is promoted to general partner. (Source: Fortune / CoinDesk)

Meta faces collective lawsuits from five major publishers over copyright infringement. Macmillan, McGraw-Hill, Elsevier, Pearson, and Springer accuse Meta of unauthorized large-scale copying of books and journal articles to train Llama. (Source: FT / The Verge)

Meta plans to beta test consumer AI agent Hatch within June. Simultaneously, it will launch AI-based shopping tools on Instagram. The trend of AI agents moving from enterprise tools to consumer applications is accelerating. (Source: FT / Reuters / 36Kr)

Click to learn about Rhythm BlockBeats job openings

Join Rhythm BlockBeats’ official community:

Telegram subscription group: https://t.me/theblockbeats

Telegram discussion group: https://t.me/BlockBeats_App

Twitter official account: https://twitter.com/BlockBeatsAsia