NVIDIA Just Made Its Biggest Purchase Ever
Спикеры
Josh Kale
— Соведущий, подкаст Limitless (Bankless)
Ejaaz Ahamadeen
— Соведущий, подкаст Limitless (Bankless)
NVIDIA's blowout quarter and Hugging Face acquisition dominated a week of AI infrastructure news, reinforcing the hosts' bullishness on NVIDIA's GPU and open-source strategy. They also covered autonomous vehicle competition between Waymo's hardware upgrade and Tesla's cheaper CyberCab, new Chinese AI silicon, robotics learning progress, and Apple's upcoming AI hardware event. The episode treats AI compute demand and robotics as accelerating, while Leopold and Anthropic stories are viewed as mostly noise.
- NVIDIA reported nearly $100B quarterly revenue and raised growth guidance.
- Hugging Face acquisition seen as a strategic open-source GPU demand driver.
- Waymo's new chip reduces hardware cost and brings autonomy on-device.
- Tesla's CyberCab undercuts Waymo on cost and scales beyond geofencing.
- Chinese inference silicon, likely Huawei, reached scale with GLM 5.3 Flash.
- Robotics in-context learning is accelerating the path to useful robots.
- Apple's September 9 event is expected to launch AI-specific hardware.
Идеи
NVIDIA's record earnings show unmatched demand
The Hugging Face acquisition deepens NVIDIA's open-source AI moat: open-weight models generate huge token volume, hundreds of niche models will run on NVIDIA GPUs, and it hedges against large customers building custom chips by ensuring more open-source demand. Open source also moves NVIDIA up the stack into model deployment and CUDA optimization.
Waymo's cheaper powerful chip improves economics
Waymo's new vehicle hardware introduces an on-device AI chip that is 20x more powerful and 5x cheaper than the prior generation, moving compute from cloud dependence into the car and cutting sensor/hardware cost to about $25,000. This improves reliability and unit economics for Alphabet's autonomous vehicle business.
Tesla CyberCab undercuts robotaxi cost
Tesla's CyberCab pairs smartphone-grade cameras with a vehicle whose total cost of goods sold may be around $25,000, similar to Waymo's sensor suite alone, and is deploying in Austin, Dallas, Houston, Miami, Orlando and Tampa. Josh argues its lower cost per mile and FSD that works without geofencing could let robotaxis scale broadly within 12 months.
Robotics learning breakthrough accelerates robot revolution
A robotics model breakthrough using in-context learning lets robots learn from a single video, generalize past human mistakes, and use arbitrary tools. Combined with paid human data collection, Josh sees this as evidence the robot revolution is approaching and that robots will increasingly automate physical tasks.
Apple event may ignite AI upgrade cycle
Apple's September 9 hardware event is expected to reveal AI-specific new devices, including iPhone 18 Pro and a foldable, under new leadership. Josh views this as the long-awaited hardware unlock for Apple Intelligence and Siri after years of waiting, creating a possible upgrade catalyst.
