One of the most underrated stories in AI right now is where the models are actually coming from. Chinese AI models have jumped from roughly 11% of weekly token usage by U.S. companies a year ago to more than 30% since February 2026 — peaking near 46%. DeepSeek is developing its own inference chip to cut its dependence on Nvidia and Huawei, and open-source infrastructure players like Together AI just raised $800 million to help enterprises run cheaper alternatives to closed models.
The center of gravity in AI is shifting, and it's happening faster than most headlines convey. Podcasts are the best way to actually follow it. Here's a listener's guide.
Why This Matters
- Open and cheap is winning share. Enterprises are routing real workloads to Chinese and open-source models because they're capable and dramatically cheaper. That's a structural shift, not a fad.
- The hardware angle. DeepSeek building its own chip is a direct challenge to Nvidia's grip and to U.S. export-control strategy.
- It reframes the "AI race." It's no longer just OpenAI vs. Anthropic vs. Google — it's closed vs. open, and U.S. vs. Chinese labs.
The Best Podcasts for the Story
- Real-time tech and macro — roundtables like All-In for the competitive and geopolitical stakes.
- Builder and investor shows — the a16z Podcast and similar for how open-source economics are actually reshaping deployment.
- China-tech specialists — shows focused on Chinese technology and policy, where you'll hear nuance the mainstream misses on DeepSeek, export controls, and domestic chips.
How to build a feed: search "DeepSeek," "open-source AI," and "China AI" across Spotify, Apple, and YouTube; mix one macro show, one builder show, and one China specialist.
What to Listen For
- Open vs. closed economics. Why running an open model can be far cheaper — and what enterprises give up (if anything).
- The chip independence play. How realistic is DeepSeek's own silicon, and what it means for Nvidia.
- Export controls. Whether U.S. restrictions are slowing China down or accelerating its self-sufficiency.
- Trust and data. The real questions companies weigh before routing workloads to Chinese models.
Don't Just Listen — Capture It
This story moves weekly and is dense with numbers and names. Listen to a sharp breakdown of the open-source shift and a week later you'll keep the vibe but lose the specifics.
- Paste the episode link into DriftNote for a structured summary — overview, key topics, takeaways, and quotes with timestamps.
- Skim it right after listening and note the figures that surprised you.
- Save it in Notion so your AI notes stay searchable as the landscape shifts.
A Fast Listening Plan
- Start with a macro roundtable for the competitive picture.
- Follow with a builder show on open-source economics.
- Finish with a China-tech specialist on chips and policy.
Summarize each as you go, and you'll understand the shift better than most people fixated on the U.S. labs alone.
Where to Go From Here
- Try the free podcast summary tool
- AI agents in 2026: the agentic shift
- Nvidia and the AI chip boom
- Notion podcast notes template
The AI map is being redrawn from the open-source edge in. Listen well, capture the reasoning, and you'll see it before the headlines catch up.