Google is in the AI driver's seat
The bull case argued for Alphabet is that the company owns the entire hardware and software stack, allowing it to train frontier models without relying on third-party chips.
The argument
The guest argued that Google's proprietary Tensor Processing Units (TPUs) allowed them to train Gemini 3 without any Nvidia chips. He asserted that Google's deep bench of PhDs, full-stack ownership, and Android distribution platform put them in a superior position compared to competitors like OpenAI who must buy expensive data center space and GPUs.
The thesis, stress-tested
✓ What validates it
- ✓Increased adoption of Gemini pro subscriptions to offset potential ad revenue declines
- ✓Licensing agreements of Gemini models to other major tech players like Apple
▸ Risks discussed
- ▸AI could cannibalize Google's core search and sponsored ads business
- ▸High valuation of mega-cap tech stocks makes them less attractive entry points today
Hear it yourself
"When it dropped, it was the best model in the world, exclusively on Google chips. So they have these chips called TPUs, tensor processing units. They have the full stack, dude. They don't even need NVIDIA. It's wild. They didn't use an NVIDIA chip to train their AI, and they have world class AI. So it's just really exciting."
00:00 / 00:24
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE