🚨NVIDIA H20 Chip Crisis: China's AI Apocalypse Unfolds!

Welcome to AI Entrepreneurs

In this issue: Nvidia’s AI chip crisis deepens in China, Anthropic cracks open Claude’s “thinking” process, ElevenLabs unveils voice-perfecting Actor Mode, Amazon’s AI shops by your passion, and AI predicts dementia in your sleep. Plus, Runway’s AI Film Festival returns and Kling 1.6 sets a new video gen benchmark.

Let’s dive into the week’s biggest AI breakthroughs.

NVIDIA

AI Chip Crunch: China's H3C Warns of Nvidia H20 Shortage

Nvidia’s H20 AI chip—the only high-end processor the U.S. permits for sale in China—is running low, warns top Chinese server maker H3C. The spike in demand is fueled by Chinese tech giants like Tencent, Alibaba, and ByteDance adopting DeepSeek’s cost-efficient AI models. H3C announced it will prioritize long-term, high-profit customers as new shipments remain uncertain due to global supply disruptions and regulatory changes.

Meanwhile, China's authorities have urged companies to use energy-efficient chips in data centers—guidelines the H20 currently fails to meet. Nvidia is reportedly modifying the chip to align with these new rules and has requested discussions with China's top regulators. As chip supply dwindles and geopolitical tensions grow, the shortage could stall China’s AI ambitions while putting further pressure on Nvidia’s $17B China market.

ANTHROPIC

Inside Claude’s Mind: Anthropic's AI Microscope Shows Poetry, Planning & Multilingual Thinking

Anthropic has unveiled an “AI microscope” that reveals how its Claude 3.5 Haiku model thinks and plans. Unlike traditional models that generate text word by word, Claude exhibits signs of forward planning—especially when composing poetry. Researchers discovered that Claude selects rhyming targets first, then constructs lines that lead to them, a process closer to human creativity than mere prediction. It also handles complex questions with multi-step logic, like linking “Dallas is in Texas” to “Austin is the capital,” suggesting real-time reasoning rather than memorization.

The study shows Claude activates a "universal concept" for translating words across languages, indicating deep understanding. Google Research found AI models like Whisper mimic some human brain patterns in language tasks, but key differences exist. AI processes large chunks at once, while the human brain handles language sequentially. These insights suggest a future where AI not only speaks fluently but also thinks in human-like ways.

ELEVENLABS

ElevenLabs Launches Actor Mode: Total Voice Control with AI Precision

ElevenLabs has just dropped one of its most anticipated features — Actor Mode — now live in the ElevenLabs Studio. This new mode lets creators narrate scripts using their own voice to control pacing, emotion, and intonation, and then instantly convert the performance into any AI voice from the library or a cloned voice. Whether you're producing podcasts, TikToks, YouTube voiceovers, or audiobook narrations, Actor Mode brings studio-level delivery to your screen in seconds.

With full creative control, Actor Mode bridges the gap between human expression and AI audio generation. It’s perfect for anyone who wants to inject drama, nuance, or energy into spoken content. Try it now on ElevenLabs Studio — new users get 10,000 free credits, and affiliates can start earning today.

WITH MINDSTREAM

Mindstream is how 200,000+ readers keep on top of AI. The daily newsletter delivers actionable insights and the news you need to know to stay ahead of the curve and maximize AI in your daily life. The best part? It’s completely free. Get smarter on AI in just a few minutes each day. Join the club.

AMAZON

Amazon's AI Gets Personal: New “Interests” Feature Shops for Your Hobbies 

Amazon has rolled out a new AI-powered feature called Interests that turns your passions into personalized product picks. Found in the “Me” tab of the Amazon Shopping app, this smart tool lets users describe what they love—like “wooden brain teaser puzzles” or “romantic home decor under $100”—and instantly generates tailored shopping suggestions. You can create multiple profiles, tweak them by style, color, or price range, and the AI keeps the ideas flowing with fresh picks.

This is just one of nearly 1,000 AI projects in the works at Amazon. “Interests” may run on a scaled version of the company’s Nova series reasoning model, while future versions could plug into Alexa+ and specialized tools like Health AI. For now, it’s rolling out to select U.S. customers, with a full release coming soon. Personalized shopping just got a serious upgrade—driven by AI.

AI HEALTH

AI Predicts Dementia in Your Sleep

AI is getting ahead of Alzheimer’s—literally. A new tool from Mass General Brigham scans sleep EEGs and predicts dementia risk five years in advance with 85% accuracy. It tracks tiny gamma shifts during deep sleep, hinting at a future where wearables detect neurodegeneration long before symptoms appear.

Hungry for More AI Healthcare Wins?
Dive into AIHealthTech Insider – Issue #41 for the full scoop on AI’s medical revolution.

Stay Ahead with Medicine’s Future—Subscribe Now!

AI BYTES

Runway’s AI Film Festival Returns – Submit by April 13!

The third annual AI Film Festival by Runway is back this June, spotlighting visionary creators at the crossroads of storytelling and AI. Finalist films (1–10 minutes, any genre) will be showcased at gala screenings in NYC (June 5) and LA (June 12), with over $60,000 in awards up for grabs. Submissions are open until April 13, and tickets are now available for both cities. Don't miss this chance to premiere your AI-driven film on the big stage!

Kling 1.6 Achieves Perfect 1000 ELO in AI Video Generation

Kling AI has made history with its latest innovation, Kling 1.6, achieving a perfect 1000 ELO score and cementing its position as the #1 tool in global AI video generation rankings. This revolutionary model boasts unparalleled advancements in prompt adherence, dynamic motion generation, and stunning visual fidelity. With 195% enhancement in image-to-video capabilities compared to its predecessor Kling 1.5, Kling 1.6 offers creators lifelike videos that elevate storytelling to new heights.

AI CREATIVITY

How to Create Cinematic AI Shots with Kling 1.6 Pro

1. Start by imagining your scene in two parts:

   Shot 1: Picture a lone man trekking through endless golden sand dunes beneath a scorching sun. His shadow trails behind him as he pauses, looking up at the blinding sky. The camera tilts upward, capturing the glare and heat shimmering off his sweat-drenched face and windblown clothes.

   Shot 2: He grabs a pair of mirrored aviator sunglasses from his collar and slides them on. The camera zooms in close — now we see the vast desert reflected in his lenses. The light softens to golden-hour hues, painting the background in cinematic warmth.

2. Now open Kling 1.6 Pro, the cutting-edge video generation tool that's leading the AI film revolution.

3. Use a natural language prompt to describe the scene above — Kling's intelligent algorithms will interpret the lighting, movement, textures, and emotion.

4. Preview your generated video and refine details like pacing, camera angles, and mood with easy tweaks.

🎬 With Kling 1.6, you're not just creating clips — you’re crafting stories.
It’s the new standard for cinematic AI content.