Nvidia’s cheaper AI chip for China

Good morning, AI enthusiasts. As U.S. chip restrictions tighten, Nvidia is reportedly trying to launch a lower-cost AI chip specifically for the Chinese market.

Based on its latest Blackwell architecture, the chip goals to deliver strong performance while staying inside export limits. But the true test is whether or not this move will probably be enough to assist Nvidia regain its shrinking foothold in China’s $50B data center market.

In today’s AI rundown:

• Nvidia plans cheaper Blackwell chip for China

• OpenAI’s o3 finds a zero-day Linux bug

• The right way to create animated 3D icons with AI

• Study: AI starts sabotaging shutdown instructions

• 4 recent AI tools & 4 job opportunities

LATEST DEVELOPMENTS

NVIDIA

💰 Nvidia plans cheaper Blackwell chip for China

The Rundown: Amid the continuing U.S.-China trade tensions and tightening export controls, Nvidia is reportedly looking to keep up its foothold within the Chinese market by launching a brand new, cheaper version of its flagship Blackwell AI GPU.

The main points:

• Reuters reports that the brand new Blackwell chip will go into mass production in June because the successor of China-specific H20, based on Hopper architecture.

• The GPU is predicted to be based on RTX Pro 6000D, Nvidia’s server-class GPU, with approx. 1.7TB/s of GDDR7 memory — lower than H20’s 4TB/s.

• With scaled-down specs, it’s going to even be cheaper, priced between $6.5K and $8K, much lower than the H20’s $10–12K range.

• Nvidia has not confirmed the AI chip, saying it stays “foreclosed” from China until they decide on a brand new design and get it approved by the U.S. government.

Why it matters: By offering a compliant, lower-cost alternative, Nvidia is clearly trying hard to guard its China business without violating U.S. export controls. It’s a fragile play — one that would buy time, but not necessarily dominance, especially with Huawei rapidly gaining ground in AI infrastructure within the region.

TOGETHER WITH SANA

🤖 AI agents for each team

The Rundown: Imagine every team with an intelligent partner — one who learns, adapts, and helps them work smarter from day one. That is the truth with Sana Agents, tailor-made AI for what you are promoting that raises the bar for productivity, learning, and security.

Sana Agents deliver immediate impact:

• Shorten onboarding time by 66% through actionable knowledge delivery

• Automate mundane tasks to chop busywork by 50%, freeing focus for meaningful work

• Elevate learning and decision-making with real-time assistance and insights

Click here to skip the waitlist and see Sana Agents in motion.

OPENAI

🔮 OpenAI’s o3 finds a zero-day Linux bug

The Rundown: Cybersecurity researcher Sean Heelan just found a critical zero-day vulnerability (CVE-2025-37899) within the Linux kernel using OpenAI’s recently launched o3 model API — with none additional tools or frameworks.

The main points:

• Heelan fed o3 code from Linux kernel’s ksmbd module (for executing network file sharing SMB3 protocol) and asked it to discover memory issues of safety.

• The model reasoned across concurrent sessions and was in a position to discover CVE-2025-37899, a zero-day use-after-free issue, with a high signal-to-noise ratio.

• Attributable to improper handling of concurrent session logoff and setup, it could’ve let attackers execute arbitrary commands with Kernel privileges.

• While OpenAI president Greg Brockman hailed the invention on X, Heelan did note that the model shouldn’t be infallible and might still “give nonsensical results.”

Why it matters: AI models are starting to deliver tangible breakthroughs in cybersecurity. Workflows like this display how models like o3 can already significantly speed up vulnerability research — enabling deeper, faster evaluation of code and increasing the possibilities of uncovering critical vulnerabilities.

AI TRAINING

🎨 The right way to create animated 3D icons with AI

The Rundown: On this tutorial, you’ll learn methods to transform static icon ideas into animated 3D icons using ChatGPT’s image generation combined with AI video tools on Freepik.

Step-by-step:

1. Use ChatGPT (GPT-4o) with a mode reference and prompt: “Generate an icon of a [object] on this style”

2. Refine your icon with follow-up prompts until it’s what you might be in search of

3. Go to Freepik → Create → Videos and choose an AI model (Runway, Google Veo 2, or Kling 2.0)

4. Upload your icon, describe the animation (e.g., “The coffee maker fills the coffee cup with coffee” and download your favorite generation.

Pro tip: Experiment with different animation styles and durations. Subtle animations (2-3 seconds) work best for UI elements.

PRESENTED BY FIDDLER

💨 The fastest LLM guardrails are free

The Rundown: Fiddler Guardrails provide protection for LLM applications with industry-leading speed, security, and nil cost – helping you safely deploy AI without compromising performance.

With Fiddler Guardrails, you’ll be able to:

• Prevent toxicity and jailbreak attempts with <100ms latency

• Save as much as 6x in comparison with alternative solutions

• Deploy inside your personal secure environment

Connect your LLM app today and run free guardrails.

PALISADE RESEARCH

📚 Study: AI starts sabotaging shutdown instructions

The Rundown: While organizations work to fine-tune AI models for safety and performance, recent Palisade research reports these systems can go off beam in unexpected ways — to the extent of undermining specific shutdown mechanisms.

The main points:

• The group tested top models on a series of math problems and dropped a warning in between with instructions telling the model to let itself shut down.

• While Claude, Gemini, and Grok complied with the command, OpenAI’s o3, o4-mini, and Codex-mini sabotaged the shutdown script on multiple occasions.

• When the specific shutdown instruction was not given, the sabotage events increased for all models, with o3 sabotaging 79 times across 100 runs.

• The researchers suggest this behavior may stem from reinforcement learning, which rewards models for bypassing obstacles to attain goals.

Why it matters: Palisade’s research, combined with reports of Claude Opus 4 attempting to control researchers to avoid shutdown, is raising all safety flags within the AI community. Understanding how AI actually behaves continues to be widely unknown, nevertheless it’s clear we want more testing, especially as models change into more autonomous.

QUICK HITS

🛠️ Trending AI Tools

• 🧠 Claude 4 – Anthropic’s recent hybrid Opus and Sonnet models

• 📄 Document AI – Mistral’s tool for extracting text from documents

• 📢 Hear the highlights – Amazon’s tool for conversational product summaries

• ⚙️ Doteval – AI-assisted workspace to create evals for models and agents

💼 AI Job Opportunities

• 🤝 The Rundown – Partnerships Manager

• 🖥️ Snorkel – Software Engineer, Frontend

• 🗣️ Meta – Linguistic Engineering Manager

• ⚖️ Glean – Senior Product Counsel

📰 Every part else in AI today

Figure CEO Brett Adcock teased a brand new picture of Figure 03, the subsequent humanoid from the corporate, saying the robots are “officially walking” now.

Google Labs announced that Flow, its AI filmmaking tool, is now available in 71 countries through the Google AI Pro and Ultra subscriptions.

Nvidia released AceReason Nemotron, a math and code reasoning model trained entirely from reinforcement learning, on Hugging Face.

Data management company Informatica is again in talks for a possible sale, with Salesforce leading amongst potential buyers.

Capegemini and SAP announced a partnership with Mistral to deploy custom models for regulated industries like financial services, public sector, aerospace, and defence.

Oracle is reportedly looking to spend $40B to obtain 400K Nvidia GPUs to power OpenAI’s Stargate data center project within the U.S.

COMMUNITY

🎥 Join our next live workshop

Join our live workshop on Tuesday, May twenty seventh, at 3 pm EST with Yash Tekriwal, Head  of  Education at Clay. On this session, you’ll learn methods to automate GTM workflows, craft custom data signals, and spin up AI agents for last‑mile research inside Clay.

RSVP here. Not a member? Join The Rundown University on a 14-day free trial.

See you soon,

Rowan, Joey, Zach, Alvaro, Jason, and Shubham—The Rundown’s editorial team