PLUS: AMD's NVIDIA rival gets Microsoft, AI spots cancer months earlier, and agents in Adobe and Blender
Happy reading
Powerful open-weight AI models are emerging from China, sparking a secret battle within the Trump administration over a potential ban. The move could protect U.S. incumbents like OpenAI, but at the cost of stifling global competition and innovation.
This sets up a clash between America's closed-model strategy and China's open-ecosystem approach. Will the U.S. prioritize protecting its AI giants, or will it allow permissionless innovation to drive the future of the technology?
In today's Next in AI:
The secret Trump battle to ban Chinese AI
AMD's NVIDIA rival gets Microsoft backing
AI spots cancer months earlier than doctors
AI agents come to Adobe and Blender
China's AI Gauntlet
Next in AI: The rapid rise of powerful, open-weight Chinese AI models is sparking a secret battle within the Trump administration over a potential ban, a move that could protect U.S. incumbents like OpenAI and Anthropic at the risk of stifling global competition.
Explained:
China is turning a hardware disadvantage into a strategic edge by releasing powerful open-weight models like Kimi K3, which rivals top U.S. systems in performance at a fraction of the cost.
U.S. officials are weighing several actions, including adding Chinese AI labs to the Commerce Department's Entity List and issuing national security advisories to discourage American companies from using the technology.
The debate pits national security hawks against pro-competition voices who argue a ban would primarily benefit a few dominant U.S. labs, a concern echoed as China's leadership appears emboldened by the progress of its open AI ecosystem.
Why It Matters: This represents a fundamental clash between America's closed-model strategy and China's open-ecosystem approach. The outcome will help decide whether the future of AI is driven by permissionless innovation or concentrated within a few protected giants.
AMD's NVIDIA Rival
Next in AI: AMD is mounting its most direct challenge to NVIDIA's market dominance with Helios, its first rack-scale AI system. The company is already hitting the ground running, securing Microsoft as a major customer alongside Meta and OpenAI.
Explained:
Major tech players including Microsoft, Meta, and OpenAI are lining up to deploy Helios in their data centers for frontier model inference and AI services.
The system is an all-in-one package, combining AMD's own GPUs, CPUs, and networking hardware in a direct challenge to NVIDIA by pairing it all with its ROCm open-source software.
As part of the partnership, Microsoft is also launching two new computing instances on Azure powered by AMD chips for agentic AI and semiconductor design.
Why It Matters: The arrival of a serious competitor in high-end AI hardware creates more choice and could eventually drive down the soaring costs of compute. This intensified competition accelerates the hardware arms race, spurring faster innovation and increasing the supply of powerful AI systems for everyone.
AI's Cancer Check
Next in AI: A breakthrough from Johns Hopkins researchers shows AI can identify early signs of pancreatic cancer on CT scans up to 16 months before human doctors, creating a massive new window for treatment.
Explained:
Pancreatic cancer is notoriously deadly because it's often found late; catching it before it spreads increases the five-year survival rate from 13% to 44%.
This isn't an isolated success—the Mayo Clinic’s REDMOD model also spots warning signs in scans 16 months prior, specifically targeting high-risk individuals.
Tech giants are also tackling this, with Alibaba’s PANDA model demonstrating 92.9% sensitivity in finding tiny lesions on routine, non-contrast CT scans.
Why It Matters: This technology gives doctors a powerful new tool to find subtle patterns that are nearly invisible to the human eye. It signals a major shift toward proactive healthcare, where AI helps detect diseases long before symptoms ever appear.
Agents in Your Apps
Next in AI: At SIGGRAPH, NVIDIA unveiled a new protocol that allows AI agents to work directly inside professional creative tools like Adobe, Blender, and Unreal Engine, automating complex tasks and streamlining workflows.
Explained:
Instead of just speeding up rendering, the Model Context Protocol (MCP) gives agents contextual awareness to handle production tasks like checking for errors, preparing files, and managing assets, all while creators maintain final control.
Major industry players are already integrating the protocol, with Adobe, Blender, and Unreal Engine leading the charge to bring agent-assisted workflows to millions of creative professionals.
The platform allows professionals to run agents locally on NVIDIA hardware, which improves responsiveness and keeps sensitive project data secure, moving beyond a total reliance on cloud services.
Why It Matters: This moves AI from a separate content generator into a collaborative partner that understands the context of your project. Professionals can now offload tedious production work to AI assistants, freeing up more time for high-level creative decisions and innovation.
AI Pulse
Alphabet plans to develop a new "Frozen v2" server chip that could run Gemini models up to ten times more efficiently by embedding parts of the architecture directly into the silicon.
NVIDIA announced its Synthetic Video Detector, a new NIM microservice designed to help media organizations identify AI-generated content by analyzing video on a frame-by-frame basis.
Bloomy launched its YC-backed, AI-powered mastery-learning platform for K-12 students, offering personalized learning paths and a Socratic AI tutor to address individual skill gaps.
Scientists warned that a rise in AI-altered images on birdwatching forums is creating fake sightings and threatening the credibility of citizen science data used in ecological research.