🗞 UAE AI Infrastructure, Claude 4 Launch, and Perplexity's Web Automation

🔎 The Latest on the AI Frontier:

• Tech Giants Unite for Massive AI Infrastructure Project in UAE

• Anthropic Unveils Claude 4 Family with Advanced Coding Capabilities

• Perplexity Develops Automated Web Browsing Task Scheduler

• AI App Builder Startup Collapses Despite Half-Billion Investment

• Apple to Open AI Development Tools to Third-Party Creators

• Other news you might find interesting

🤝 Major tech giants including G42, OpenAI, Oracle, NVIDIA, SoftBank, and Cisco announce partnership to build Stargate UAE, a next-generation AI infrastructure cluster in Abu Dhabi.

• The 1-gigawatt compute cluster will be housed in a massive 5-gigawatt UAE–US AI Campus spanning 10 square miles, representing the largest AI deployment outside the US with the first 200-megawatt phase going live in 2026.

• This marks the first major milestone in OpenAI's "OpenAI for Countries" initiative, establishing the world's first Stargate facility outside the US as part of the new US-UAE AI Acceleration Partnership framework.

• The facility will be powered by nuclear, solar, and natural gas to minimize carbon emissions while providing nation-scale compute and low-latency inferencing to accelerate scientific discovery across healthcare, energy, finance, and transportation sectors.

🚀 Anthropic launches Claude 4 family featuring Claude Opus 4 as "the world's best coding model" and Claude Sonnet 4 with significant upgrades, both offering hybrid modes for instant responses and extended thinking.

• Claude Opus 4 leads on SWE-bench at 72.5% and Terminal-bench at 43.2%, capable of sustained performance on complex tasks running for several hours, while Sonnet 4 achieves 72.7% on SWE-bench and will power GitHub Copilot's new coding agent.

• New capabilities include extended thinking with tool use, parallel tool execution, improved memory through local file access, and Claude Code now generally available with VS Code and JetBrains integrations plus GitHub Actions support for background tasks.

• Pricing remains unchanged at $15/$75 per million tokens for Opus 4 and $3/$15 for Sonnet 4, with both models available on Anthropic API, Amazon Bedrock, and Google Cloud's Vertex AI, plus four new API capabilities including code execution and Files API for developers.

🤖 Perplexity is developing a "Scheduled Tasks" feature to automate web browsing routines, found in their codebase but not yet active in the user interface.

• The feature will work through Perplexity's web client but is expected to be more effective on their Comet browser, which can interact deeply with websites to automate tasks like LinkedIn job applications, social media posting, and content publishing on platforms like Ghost.

• Unlike ChatGPT's scheduled actions that only send email links, Perplexity's implementation may include actual content delivery in emails, offering users automated daily news digests and recurring summaries with fuller information.

• The browser-agnostic automation positions Comet as a more capable browser-agent hybrid compared to tools like Manos, with potential for hands-free execution of routine tasks through prompt chaining and cloud-based workflows.

💸 Builder.ai, a British startup that pitched AI-powered app development, has filed for bankruptcy despite raising over $500 million from investors including Microsoft and Qatar's sovereign wealth fund.

• The company, previously valued near $1 billion, was exposed in 2019 for using human engineers rather than AI for most coding work, echoing the historical "Mechanical Turk" chess-playing automaton that secretly relied on hidden human operators.

• Microsoft CEO Satya Nadella recently claimed 30% of code in some repositories is AI-written, but GitHub threads show developers spending extensive time correcting AI coding mistakes that would embarrass junior programmers, highlighting the gap between AI promises and reality.

• Builder.ai's collapse from financial mismanagement and over-optimistic forecasts serves as a cautionary tale for the generative AI coding industry, demonstrating that current AI tools remain far from replacing human developers despite significant investment and hype.

🍎 Apple plans to release a software development kit in iOS 19 that will allow developers to use Apple's AI models in their apps, making the same technology powering Apple Intelligence features available to third-party creators.

• The SDK will initially focus on smaller on-device models used for notification summaries, Writing Tools, Genmoji, and Image Playground, as Apple hopes opening up its AI framework will drive better consumer adoption of AI features that haven't seen widespread use.

• Currently developers can only integrate existing Apple Intelligence features into apps but cannot create new AI capabilities using Apple's framework, forcing them to rely on third-party models like ChatGPT or Claude instead.

• The announcement will come at WWDC 2025 on June 9, while Apple continues developing a more advanced LLM-powered version of Siri comparable to ChatGPT that isn't expected until iOS 20 in 2026 at the earliest.

More news you might find interesting:

• Time Magazine accidentally published and quickly removed an embargoed story confirming Anthropic's new Claude model announcement, with the article later scraped and shared on Newsbreak before being taken offline.

• Musk's DOGE team is expanding use of his Grok AI chatbot across U.S. federal agencies to analyze government data, raising conflict-of-interest concerns.

• First Lady Melania Trump releases an AI-generated audiobook of her memoir, calling it "the future of publishing" with her voice entirely recreated using artificial intelligence technology.

• Enchant launches a zero-equity accelerator for gaming and AI startups, founded by veterans from Sony, Supercell, EA, Ubisoft and Playrix, allowing founders to keep 100% of their cap table while receiving intensive support.

• Google I/O 2025 showcases major AI advancements including Gemini 2.5 improvements, new subscription tiers, and Android XR smart glasses set to launch later this year.

Have any feedback? Send us an email