What are the key takeaways from this a16z Podcast episode?

Key insights include: **Infrastructure ownership vs. utility:** The fiber optic cables laid during the 1990s Internet boom remained useful for decades even after their original owners went bankrupt. Today's GPU infrastructure follows the same pattern — Volpi predicts a market correction will reshuffle ownership among the 200+ Neo Clouds, many with weak balance sheets, while the underlying compute assets retain long-term value regardless of who holds them.; **GPU asset life extension:** H100 GPUs originally modeled with a five-year useful life now rent at $2.50–$3.00 per GPU hour, higher than their original $1.82 rate, because AI-driven demand outpaces supply. Investors and operators should underwrite GPU assets on extended timelines rather than conservative depreciation schedules, as scarcity sustains pricing power across older hardware generations.; **Cisco M&A playbook for NVIDIA:** Cisco's M&A team of fewer than 10 people executed roughly 70 acquisitions, generating 50% of Cisco's $30B revenue from acquired sources. The key was accepting a one-third failure rate, using all-equity deals, and maintaining a centralized integration process. Volpi advises startup founders to build direct relationships with potential acquirers through regular coffees rather than waiting for inbound interest.

What did Mike Volpi discuss on a16z Podcast?

Former Cisco executive Mike Volpi, now at Hanabi Capital, draws parallels between the 1990s fiber optic Internet buildout and today's AI infrastructure boom, analyzing GPU scarcity, NVIDIA's M&A strategy, networking architecture shifts, and why the underlying compute assets will outlast any coming market correction. Key topics include: **Infrastructure ownership vs. utility:** The fiber optic cables laid during the 1990s Internet boom remained useful for decades even after their original owners went bankrupt. Today's GPU infrastructure follows the same pattern — Volpi predicts a market correction will reshuffle ownership among the 200+ Neo Clouds, many with weak balance sheets, while the underlying compute assets retain long-term value regardless of who holds them.; **GPU asset life extension:** H100 GPUs originally modeled with a five-year useful life now rent at $2.50–$3.00 per GPU hour, higher than their original $1.82 rate, because AI-driven demand outpaces supply. Investors and operators should underwrite GPU assets on extended timelines rather than conservative depreciation schedules, as scarcity sustains pricing power across older hardware generations..

How long is this episode of a16z Podcast?

This episode is 56 minutes long. SignalCast provides an AI-generated summary so you can get the key insights in about 3 minutes.

a16z Podcast

What the Internet Boom Teaches Us About the AI Buildout | Mike Volpi

November 11, 2026

56 min episode · 2 min read

Mike Volpi

Episode

56 min

Read time

2 min

Topics

Relationships, Investing, Startups

AI-Generated Summary

Published Nov 11, 2026

Key Takeaways

✓Infrastructure ownership vs. utility: The fiber optic cables laid during the 1990s Internet boom remained useful for decades even after their original owners went bankrupt. Today's GPU infrastructure follows the same pattern — Volpi predicts a market correction will reshuffle ownership among the 200+ Neo Clouds, many with weak balance sheets, while the underlying compute assets retain long-term value regardless of who holds them.
✓GPU asset life extension: H100 GPUs originally modeled with a five-year useful life now rent at $2.50–$3.00 per GPU hour, higher than their original $1.82 rate, because AI-driven demand outpaces supply. Investors and operators should underwrite GPU assets on extended timelines rather than conservative depreciation schedules, as scarcity sustains pricing power across older hardware generations.
✓Cisco M&A playbook for NVIDIA: Cisco's M&A team of fewer than 10 people executed roughly 70 acquisitions, generating 50% of Cisco's $30B revenue from acquired sources. The key was accepting a one-third failure rate, using all-equity deals, and maintaining a centralized integration process. Volpi advises startup founders to build direct relationships with potential acquirers through regular coffees rather than waiting for inbound interest.
✓Networking architecture disruption: AI data centers run fully utilized, predictable traffic flows — the opposite of the stochastic, statistically multiplexed patterns that Broadcom's Tomahawk switch architecture was designed for. A new networking architecture optimized for constant, deterministic flows (circuit-based rather than packet-based) could dramatically reduce data center networking costs and represents a significant opportunity for new entrants unencumbered by legacy product lines.
✓Silicon scarcity governs the AI cycle: Fab capacity, not power or capital, is the binding constraint on AI infrastructure supply for the next two to five years. Chip architectures with 3D DRAM and hybrid copper bonding address the dominant bottleneck — memory bandwidth for KV cache-intensive reasoning workloads. Founders building custom silicon should align chip design with specific model architectures rather than targeting general-purpose compute.

What It Covers

Former Cisco executive Mike Volpi, now at Hanabi Capital, draws parallels between the 1990s fiber optic Internet buildout and today's AI infrastructure boom, analyzing GPU scarcity, NVIDIA's M&A strategy, networking architecture shifts, and why the underlying compute assets will outlast any coming market correction.

Key Questions Answered

•Infrastructure ownership vs. utility: The fiber optic cables laid during the 1990s Internet boom remained useful for decades even after their original owners went bankrupt. Today's GPU infrastructure follows the same pattern — Volpi predicts a market correction will reshuffle ownership among the 200+ Neo Clouds, many with weak balance sheets, while the underlying compute assets retain long-term value regardless of who holds them.
•GPU asset life extension: H100 GPUs originally modeled with a five-year useful life now rent at $2.50–$3.00 per GPU hour, higher than their original $1.82 rate, because AI-driven demand outpaces supply. Investors and operators should underwrite GPU assets on extended timelines rather than conservative depreciation schedules, as scarcity sustains pricing power across older hardware generations.
•Cisco M&A playbook for NVIDIA: Cisco's M&A team of fewer than 10 people executed roughly 70 acquisitions, generating 50% of Cisco's $30B revenue from acquired sources. The key was accepting a one-third failure rate, using all-equity deals, and maintaining a centralized integration process. Volpi advises startup founders to build direct relationships with potential acquirers through regular coffees rather than waiting for inbound interest.
•Networking architecture disruption: AI data centers run fully utilized, predictable traffic flows — the opposite of the stochastic, statistically multiplexed patterns that Broadcom's Tomahawk switch architecture was designed for. A new networking architecture optimized for constant, deterministic flows (circuit-based rather than packet-based) could dramatically reduce data center networking costs and represents a significant opportunity for new entrants unencumbered by legacy product lines.
•Silicon scarcity governs the AI cycle: Fab capacity, not power or capital, is the binding constraint on AI infrastructure supply for the next two to five years. Chip architectures with 3D DRAM and hybrid copper bonding address the dominant bottleneck — memory bandwidth for KV cache-intensive reasoning workloads. Founders building custom silicon should align chip design with specific model architectures rather than targeting general-purpose compute.

Notable Moment

Volpi recounts how Cisco, after never missing a quarterly earnings target across a decade as the world's most valuable company, watched demand collapse so suddenly in early 2001 that three consecutive weeks of zero January recovery signaled a 35–40% revenue miss, sending the stock from roughly $80 to $9.

Know someone who'd find this useful?

Episode Transcript

We live in a world where demand for AI capacity tokens is near infinite, and supply is incredibly constrained. So another way to put it is we all just live in Jensen's world. You were doing back then M and A for the most valuable company on the planet. The retro on the M and A process that I think worked well for Cisco was we were not afraid to fail. And we did it with high enough frequency that we got a few outliers. You bought this thing for $500,000,000 and, like, it's doing a billion in earnings two years later. When you see the AI build out, how do you look back on your previous experiences? If you look at the revenue ramp of Anthropic, OpenAI are definitely experiencing growth rates that were far superior to what we saw at the time. Do I think the build out that's happening now is useful and will be fully utilized, if not more? A 100%. Will it be owned by the same people that own it today? The last great infrastructure boom built the Internet. This one is building the infrastructure for intelligence. What can the first tell us about the second? In this episode, a 16 z's Raguragurag and Guino Abbotseller sit down with Hanabi Capital general partner and former Cisco executive Mike Volpe. They compare today's investment in GPUs and data centers with the fiber boom of the late nineteen nineties. And why Mike thinks the infrastructure will remain valuable even if a correction reshuffles who owns it. They also discuss NVIDIA, custom AI chips, networking, and what Cisco's acquisition strategy can teach founders and investors today. Mike's view, the opportunity underneath this build out could ultimately be even larger than the Internet boom. Great to have you here on the podcast, Mike, and it's a very, very opportune time to have because you were one of the architects of the great Internet build out when you were at Cisco. And now, of course, you've got the great AI build out happening. So we don't... I and the audience would love to get your thoughts on then and now from all different dimensions, from the actual build out, the capitalization of it, the technology differences, etcetera, etcetera. So let's start for those that don't know or remember. Mike was a senior exec at Cisco overseeing a lot of the M and A and strategic moves that Cisco made in the late nineties. And of course, that was the time the Internet got built out. So Mike was in the middle of all of it. So clearly, let's start from there. So you were there when the Internet started taking off, and then you guys rushed into a lot of great things, and the industry rushed to build a lot of bandwidth, etcetera, etcetera, which for some time was not useful, and then it turned out it was actually too little. Right? And there are some loose similarities to …

Get the full transcript (9,504 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all a16z Podcast transcripts →