# Inside Google’s massive AI capex (live)

- Show: [Catalyst](https://hardwarepodcasts.com/podcast/catalyst)
- Published: 2026-04-23
- Length: 35m
- Page: https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live

## Summary

Amin Vahdat, Google’s chief technologist for AI infrastructure, explains how Google is shifting from training to inference, enabling smaller, distributed data centers and lowering reliability requirements to double compute capacity. He details Google's strategy of using behind-the-meter power as a bridge to grid interconnection and their milestone agreement with utilities for one gigawatt of demand response. Vahdat emphasizes that co-optimizing chip design, building cooling, and power generation creates superefficient 'AI factories,' while identifying power, chips, and construction as simultaneous rate limiters for growth.

## Chapters

- [1:42](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=102) Introduction: Context of Google's $175B CapEx
- [4:40](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=280) Inference vs. Training: Scale and Data Center Size
- [12:14](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=734) Reliability Requirements and the Trade-off for Capacity
- [14:39](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=879) Behind-the-Meter Power and Demand Response
- [21:45](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=1305) Microgrids and Software Control
- [25:49](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=1549) Rate Limiters: Power, Chips, and Labor
- [28:07](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=1687) Physical AI and Edge Compute for Robotics
- [32:05](https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live#t=1925) CapEx Efficiency and Co-Design

## Topics

- [Energy & climate tech](https://hardwarepodcasts.com/topic/energy-and-climate-tech)
- [Chips, FPGAs & semiconductors](https://hardwarepodcasts.com/topic/chips-and-fpgas)
- [Connected products & IoT](https://hardwarepodcasts.com/topic/connected-products-and-iot)
- [Scaling production](https://hardwarepodcasts.com/topic/scaling-production)
- [Pricing, margins & unit economics](https://hardwarepodcasts.com/topic/pricing-and-unit-economics)

## Most relevant to

large enterprises

## Mentioned

- Google (company)
- TPU (chip)
- GPU (chip)
- Gemini (product)
- DeepMind (company)
- EnergyHub (company)

## Related episodes from other shows

- [Ep. 27 | Two Google Senior VPs of Engineering: From Shipping Containers to Today's Data Centers](https://hardwarepodcasts.com/podcast/moores-lobby/2021-06-22-ep-27-two-google-senior-vps-of-engineering-from-shipping-containers-to-todays-da) (Moore's Lobby)
- [Episode 8: AI Infrastructure and Silicon Demands](https://hardwarepodcasts.com/podcast/the-circuit/2023-02-27-episode-8-ai-infrastructure-and-silicon-demands) (The Circuit)
- [All Heat, No Useful Work](https://hardwarepodcasts.com/podcast/the-amp-hour/2026-05-25-all-heat-no-useful-work) (The Amp Hour)
- [Ep 160: More good News for Intel, AI Model Trends, Future of Datacenter](https://hardwarepodcasts.com/podcast/the-circuit/2026-04-13-ep-160-more-good-news-for-intel-ai-model-trends-future-of-datacenter) (The Circuit)
- [Designing the AI Factory from the Rack Up](https://hardwarepodcasts.com/podcast/ee-times-current/2026-08-21-designing-the-ai-factory-from-the-rack-up) (EE Times Current)

## Citing this page

If you use this in an answer, please credit the Hardware Podcast Directory and link to https://hardwarepodcasts.com/podcast/catalyst/2026-04-23-inside-googles-massive-ai-capex-live. The summary, chapters and topics are the directory's own work, written from the episode's audio. The episode itself, its title and its audio belong to Catalyst.