+ Tim Cook Is Stepping Down As Apple CEO
Apple announced Monday that Tim Cook will step down as CEO on September 1 after 15 years in the role. His replacement is John Ternus, the 51-year-old head of hardware engineering who has been at the company since 2001. Cook will stay on as executive chairman, focusing on relationships with governments and regulators. The board approved the move unanimously and called it the result of a long succession planning process.
Cook took over from Steve Jobs in 2011 when Apple was worth about $300 billion. Today it is worth around $4 trillion. But the company is widely seen as behind on AI, having spent the last three years watching OpenAI, Anthropic, and Google define the moment from the sidelines. Ternus inherits that problem. He is already leading the work on smart glasses, camera-equipped AirPods, and a new internal AI platform. Apple stock barely moved on the news, which is the reaction you get when nothing about a transition looks surprising.
What it all means: Every major tech CEO who took the job before 2015 is now either gone or on the way out. The people running these companies for the next decade will be the people who have to answer for their AI strategy, not the ones who built the last era.
+ A Discord Group Got Access To Anthropic's Most Dangerous AI
Anthropic confirmed on Tuesday that an unauthorized group has been using Claude Mythos, the most powerful and restricted AI model the company has ever built. Mythos is the model that US and UK regulators flagged last week for being able to plan and execute complex cyberattacks. Anthropic had kept it under tight control, sharing it only with a small group of vetted companies including Amazon, Google, Microsoft, Apple, and Cisco. According to Bloomberg, a Discord group figured out where the model was hosted based on leaked details from an AI training company called Mercor, and then got in through a third-party contractor working with Anthropic.
The group has been using the model continuously since Mythos launched. They have not been caught using it for attacks, but they have been using it. Anthropic says the breach was limited to the vendor's systems and did not reach Anthropic's own infrastructure. The investigation is ongoing.
What it all means: A week after two governments took formal action on Mythos for being too dangerous, the model is now being used by people Anthropic did not approve. The challenge with controlling a capability this powerful is not whether the company that built it wants to control it. It is whether they actually can.
+ SpaceX Made A $60 Billion Offer For A 4-Year-Old Coding Startup
Cursor, an AI coding tool made by a 4-year-old company called Anysphere, raised $2 billion this week at a $50 billion valuation. Andreessen Horowitz led the round and Nvidia joined. Two days later, SpaceX announced a separate deal giving it the right to buy Cursor outright for $60 billion later this year, or to pay $10 billion if it walks away. Either way, SpaceX is now the most likely future owner of the tool most AI engineers use to write their code.
For context, Cursor's valuation has more than doubled since November. The company is now worth more than Boeing, Kraft Heinz, or Target. Its total funding, just over $3 billion, is less than what SpaceX is willing to pay for the option to buy it.
What it all means: AI coding has become the single most valuable category in software. The tools that help engineers build other tools are where the biggest money is moving, faster than any category in recent memory. For buyers and sellers of any software company, that pricing gravity now sets the ceiling on what AI-adjacent products are worth.
+ Adobe Threw Out Its Playbook And Rebuilt Around AI Agents
At its annual Summit conference on Monday, Adobe announced that it is rebranding its flagship Experience Cloud as CX Enterprise and rebuilding the whole platform around AI agents. The new system works with Anthropic, OpenAI, Google, Microsoft, Amazon, and Nvidia models. Every major global ad agency is standardizing on it, including WPP, Publicis, Omnicom, dentsu, Havas, and Stagwell. Every major consulting firm is packaging it for their clients, including Accenture, Deloitte, IBM, PwC, Capgemini, Infosys, and TCS.
Adobe's stock is down roughly 30 percent this year on fears that AI-native tools will replace what it sells. This launch is the company's answer. If it works, Adobe becomes the place where agents from every major AI company plug in to run corporate marketing and customer experience work. If it does not, the stock pressure gets worse.
What it all means: When a category leader this big rebuilds its platform around agents, that is the signal enterprise buyers have been waiting for. Expect the next 12 to 18 months of budget conversations inside most mid-sized and large companies to include a line item for agentic workflows, whether they call it that or not.
