Amazon Commits Up to $25 Billion More to Anthropic in Expanded AI Pact

Victory Giant Technology, a printed circuit board maker that supplies Nvidia, is set to begin trading in Hong Kong following a share sale that Bloomberg reports is the city's biggest listing of 2026 to date. The Shenzhen-listed company is pursuing a dual listing to tap international capital as demand for AI server components accelerates. The deal adds to a pipeline of AI-hardware suppliers seeking offshore capital, with the South China Morning Post reporting the company earlier filed for a raise of around $1 billion. The debut offers a read on investor appetite for second-tier names in Nvidia's supply chain.

Amazon announced a new investment of up to $25 billion in Anthropic, structured as an immediate $5 billion commitment plus as much as $20 billion tied to unspecified milestones, according to Engadget and multiple corroborating outlets. In exchange, Anthropic has committed to spend more than $100 billion on AWS technology over the coming decade and to continue using Amazon's Trainium silicon, securing up to 5 gigawatts of chip capacity. The deal follows two prior $4 billion Amazon investments in 2023 and 2024. Claude will also be made available natively inside the AWS portal for Amazon cloud customers.

Researchers affiliated with Berkeley AI Research (BAIR) have published a blog post describing GRASP, a gradient-based planning method designed to make long-horizon planning with learned world models more robust. According to the post, GRASP lifts trajectories into virtual states to parallelize optimization across time, injects stochasticity into state iterates for exploration, and reshapes gradients so that action signals remain clean while avoiding gradients through high-dimensional vision models. The authors list Mike Rabbat, Aditi Krishnapriyan, Yann LeCun, and Amir Bar as collaborators. The post frames long-horizon planning as the stress test where current world-model-based control methods tend to break down.

Apple announced that Tim Cook will step down as chief executive on September 1 after nearly 15 years, with senior vice president of hardware engineering John Ternus named as the company's fourth CEO. Cook will transition to executive chairman, according to a statement on the Apple Newsroom. The company's market capitalization grew from $348 billion to roughly $4 trillion during Cook's tenure, and annual revenue quadrupled to $416 billion. Ternus, 50, inherits a thinned executive bench following the departures of Apple's COO, general counsel, AI chief, CFO, and head of design, alongside strategic questions about Apple's competitive position in generative AI.

Anthropic announced Claude Design, a research preview that lets subscribers use Claude to generate presentations, prototypes and slides, according to Engadget. The app is powered by a system Anthropic calls Opus 4.7 and is available on Pro, Max, Team and Enterprise plans. It arrives the same week Adobe and Canva unveiled their own visual AI assistants.

China's AI model developers are unlikely to displace the country's enterprise software vendors because they lack the industry-specific expertise needed to serve corporate customers, according to HSBC analyst Yiran Liu in comments reported by the South China Morning Post. Liu argues that China's underdeveloped software-as-a-service market creates conditions distinct from the US, where an "AI kills software" narrative has gained traction. Instead, Liu expects model companies and legacy software firms to serve Chinese enterprises jointly. The thesis runs counter to Western market assumptions that foundation model providers will absorb application-layer value, and frames China's SaaS gap as a structural advantage for incumbents.

Apple announced that Tim Cook will transition to Executive Chairman and John Ternus, the company's senior vice president of Hardware Engineering, will become chief executive, according to an Apple Newsroom release dated April 20, 2026. Cook has served as CEO since August 2011, when he succeeded Steve Jobs. Ternus, a two-decade Apple veteran, has overseen hardware engineering for iPhone, iPad, Mac, and Apple Vision Pro. The succession arrives as Apple faces scrutiny over the pace of its artificial intelligence rollout, with Apple Intelligence features still staging across the product line and the company trailing peers on generative AI commercialization.

Bloomberg reported on April 20, 2026 that Amazon is preparing to invest an additional $5 billion in Anthropic, extending a partnership that has already positioned the AI developer as one of the largest customers of Amazon Web Services. The new tranche would build on prior Amazon commitments to Anthropic disclosed across 2023 and 2024, and arrives alongside reporting from Data Center Knowledge describing a linked $100 billion AWS compute commitment. Neither company has issued a press release confirming the figure. The reporting matters because it extends the capital and compute coupling between a hyperscaler and a frontier model developer at a scale that continues to reshape cloud economics.

Apple has named hardware engineering chief John Ternus as its next chief executive, with Tim Cook transitioning to the role of executive chairman, according to Bloomberg, which broke the story on April 20, 2026. Reuters, CNBC, The Verge, and Computerworld have confirmed the succession plan. Ternus, who has led Apple's hardware engineering organization and served as the public face of several recent product launches, becomes the fourth CEO in Apple's history. The transition arrives as Apple works to close a perceived gap with rivals on generative AI, a strategic challenge that now falls to Ternus.

Google is building out a custom AI chip supply chain with four design partners — Broadcom, MediaTek, Marvell, and Intel — and a roadmap extending to TSMC's 2-nanometre process in late 2027, according to reporting from Bloomberg and The Next Web ahead of Google Cloud Next. The seventh-generation Ironwood TPU, Google's first inference-specific chip, is now generally available on Google Cloud, with Anthropic committing to up to one million TPUs. The next-generation split assigns Broadcom's "Sunfish" to training workloads and MediaTek's "Zebrafish" to inference, with Marvell in talks for a memory processing unit.

Anthropic introduced Claude Design, an experimental product that generates visuals such as prototypes, slides, and one-pagers from text prompts, according to TechCrunch. The company says it targets founders and product managers without design backgrounds. Users describe what they want and can refine outputs through direct edits or follow-up requests.

Elon Musk did not appear on Monday for a summons from Paris prosecutors investigating his social media company X, according to The New York Times, which was corroborated by reporting from Reuters, The Guardian, the Associated Press, France 24 and Euronews. The Paris prosecutor's cybercrime division has been investigating X since January 2025 over seven potential counts under French law, including complicity in distributing child sexual abuse imagery, denial of crimes against humanity, and fraudulent data extraction. The meeting had been scheduled in February after French police raided X's Paris offices. Musk has previously called the investigation a "political attack."