PLUS: Apple's $50,000 Mac for AI and the 'forever layoff' trend
Happy reading
A celebrated mathematician is bringing old, abandoned projects back to life with the help of an AI coding agent. In just a few hours, the AI managed to modernize 25-year-old code and even build complex tools the mathematician had previously given up on.
This isn't just about making coding easier; it's about giving domain experts the power to build custom software without needing a developer's background. As these agents become more capable, what does it mean for the future of specialized software development and expert-level problem-solving?
In today’s Next in AI:
A mathematician's AI coding agent
Apple's planned $50,000 AI Mac
A new visual tool for AI code agents
The tech industry's 'forever layoff' trend
A Genius and His Agent
Next in AI: Fields Medal-winning mathematician Terence Tao is using an AI coding agent to resurrect 25-year-old code and build complex new visualization tools, turning decades-old ideas into reality in just a few hours.
Explained:
In a matter of hours, the agent successfully ported two dozen applets from defunct 1999 Java to modern JavaScript, even adding graphical upgrades to some.
Tao used a process he calls "vibe coding" to create entirely new apps, including a special relativity visualizer he abandoned 25 years ago due to its complexity.
The AI-generated code resulted in a net improvement in quality, with the agent discovering two bugs in Tao's original work while only introducing one minor bug itself.
Why It Matters: This highlights a shift from AI as a simple tool to a collaborative partner for high-level experts. It shows how agents can empower domain specialists to quickly build and prototype custom software without being professional developers.
Apple's $50,000 AI Mac
Next in AI: Apple is accelerating its chip development specifically for AI, planning a future M7 Ultra chip that could support a massive 1.5TB of memory and lead to a professional Mac with a mind-bending price tag.
Explained:
To prioritize AI, Apple is reportedly skipping high-end M6 chips like the Pro, Max, and Ultra. This move fast-tracks the M7 generation, which is being designed from the ground up for artificial intelligence workloads.
The flagship M7 Ultra, expected in 2028, is engineered to support a staggering 1.5TB of unified memory. Its performance goal is to approach the class of dedicated AI accelerators like Nvidia’s Blackwell series.
This extreme memory capacity is the key driver of the potential cost. For context, a maxed-out 2019 Mac Pro with the same amount of RAM could cost over $53,000, and that was before recent inflation and memory shortages.
Why It Matters: This strategy shows AI is no longer just another feature but the central organizing principle for Apple's entire silicon roadmap. Future Mac chip designs and release schedules are now shaped first and foremost by the immense demands of AI.
A GPS for Your AI Coder
Next in AI: A new open-source tool called mindwalk gives developers a 3D replay of their AI coding agent's every move. It turns complex session logs into an interactive map showing precisely how an agent explores and modifies your codebase.
Explained:
It visualizes your repository as a "night map" where files glow based on the agent's interaction—moss green for seen, moon white for read, and warm amber for edited.
A playback deck lets you scrub through the agent’s session, with a timeline that uses cool colors for observation (like searching) and warm colors for mutation (like editing), so key actions stand out.
The entire tool runs fully locally as a single Go binary, ensuring that no sensitive code or session data ever leaves your machine.
Why It Matters: This tool addresses a major gap in AI observability by showing the 'how' behind an agent's work, not just the 'what.' It transforms debugging from reading raw logs into a visual process, making it far easier to trust and verify an AI's work.
The AI 'Forever Layoff'
Next in AI: The tech industry is adopting a new corporate strategy of "continuous tuning," where recurring layoffs are becoming standard practice to fund AI investments, even for profitable companies.
Explained:
Tech giants like Microsoft, Cisco, and Cloudflare are cutting thousands of jobs while reporting record revenues, reframing the cuts as necessary shifts to prioritize long-term AI potential.
The trend is widespread, with AI cited as the top reason for eliminating over 120,000 roles in the tech sector so far this year.
Experts caution this constant churn can backfire by eroding institutional knowledge, driving away top talent, and incurring high costs from severance and rehiring.
Why It Matters: This marks a fundamental shift where job security in tech is no longer tied solely to company performance. For professionals, this environment makes adapting to and leveraging new AI tools a critical component of career resilience.
AI Pulse
Google reports that 75% of its code is now written by AI agents, signaling a fundamental change in how software is developed at one of the world's largest tech companies.
A survey finds that 69% of Americans support forcing large AI firms to transfer 50% of their stock into a public sovereign wealth fund to distribute AI-driven gains.
Geohot shared his success using the opencode coding agent with a local GLM-5.2 model, praising the practical power of modern AI tools for complex system administration tasks.