20VC: Anthropic compute deal Elon + $200BN Google
Category: Expert Interviews · Duration: 79 min · ▶ Watch
Speakers: Harry Stebbings
Segments (13)
- 00:00:00 · Introduction & Headlines
- A fast-paced overview of the week’s major tech news, including Anthropic’s partnerships, Cerebras’s IPO, and public market performance.
- 01:11:30 · Anthropic’s Control Over Secondary Sales
- The hosts discuss Anthropic’s recent announcement requiring board approval for all secondary sales and SPVs, and its impact on the market.
- 02:04:15 · SPVs and Secondary Market Mechanics Explained
- Rory O’Driscoll provides a detailed breakdown of how Special Purpose Vehicles (SPVs) and secondary share transfers work in private markets.
- 03:56:40 · Anthropic’s Deal with Elon Musk’s xAI
- Analysis of the partnership where Anthropic will use xAI’s compute capacity, highlighting market consolidation and strategic alignments.
- 05:22:22 · Anthropic’s $200B Google Cloud Commitment
- The discussion turns to Anthropic’s massive five-year, $200 billion commitment to Google Cloud, illustrating the scale of AI infrastructure needs.
- 06:08:15 · Goldman Sachs Predicts 24x Token Consumption Growth
- The hosts react to a Goldman Sachs report predicting a 24x increase in token consumption by 2030, driven by AI agents.
- 07:13:50 · Public Market Reactions: HubSpot, AppLovin, Cloudflare
- An analysis of why several public SaaS companies saw their stocks crash despite posting strong numbers, focusing on deceleration and AI narratives.
- 08:12:40 · The Case of ZoomInfo vs. Clay
- Jason Lemkin presents ZoomInfo as a case study of a company whose growth was ‘stolen’ by newer, AI-native competitors like Clay.
- 08:44:50 · Cerebras’s Hotly Anticipated IPO
- The team discusses the massive demand for the Cerebras IPO, its valuation, and its position as an Nvidia challenger.
- 10:00:22 · Ramp’s $40B Valuation and Fintech Dynamics
- A look at Ramp’s high valuation and the challenging economics of corporate card businesses, contrasted with the failure of competitor Parker.
- 11:15:50 · Musk vs. Altman Lawsuit Update
- A brief update on the ongoing legal battle between Elon Musk and OpenAI, including new revelations from the proceedings.
- 12:02:00 · The Irrelevance of Conference Speakers
- Jason Lemkin argues that podcasts have made traditional conference speakers and fireside chats obsolete, changing the format of SaaStr Annual.
- 12:34:30 · The Personal Cost of Founder Success
- A deep dive into the immense personal sacrifice and permanent psychological changes required to build a multi-billion dollar company.
Specific Prices (12)
| Timestamp | Item | Value | Context |
|---|---|---|---|
| 00:11:00 | Anthropic’s commitment to Google | $200 billion | A five-year commitment for cloud services. |
| 00:18:00 | Ramp’s new valuation | $40 billion | The valuation Ramp is targeting in its new fundraise. |
| 01:51:00 | Anthropic’s valuation range | $200 to $400 billion | Reported valuation in secondary markets. |
| 03:34:00 | Potential revenue from xAI deal for Anthropic | $4-5 million per year | Estimated revenue slug from selling compute capacity. |
| 03:39:00 | SpaceX revenue run rate | ~$20 billion | Total revenue run rate for SpaceX, providing context for the new revenue stream. |
| 03:48:00 | Ilya Sutskever’s net worth | $7 billion | Mentioned during the Musk vs. Altman discussion. |
| 04:55:00 | Cerebras IPO offering size | $4.8 billion | The amount the offering will raise. |
| 04:57:00 | Cerebras IPO valuation | $48 billion | The fully diluted valuation of the company at IPO. |
| 04:51:00 | Cerebras IPO price range | $150 to $160 per share | The updated price range, bumped from an initial $115-$125. |
| 08:14:00 | Gusto revenue | $1 billion | Gusto has passed the $1 billion revenue milestone. |
| 10:20:00 | Harvey legal AI software cost | $150,000 per year | Average cost per law firm for the specialized legal AI tool. |
| 11:24:00 | Jason Lemkin’s first startup exit | $50 million | Sold after 12.5 months. |
Bottleneck Claims (2)
- [00:00:45] There may not be enough developers in the solar system to keep Anthropic on its unprecedented growth path.
- Evidence: This is a rhetorical question highlighting the immense demand for talent to fuel the growth of top AI companies.
- [04:41:30] Anthropic’s biggest challenge was a lack of compute capacity to launch its new models.
- Evidence: This is the context for why Anthropic is making large-scale deals with xAI and Google to secure the necessary infrastructure.
Predictions (5)
- [00:35:00, Medium-term (next few years)] Software categories that are not compatible with an agentic workflow will enter a terminal state of decay.
- [06:09:00, By 2030] AI agents will drive a 24x increase in token consumption.
- [07:11:00, Short-term (IPO day)] The Cerebras IPO will perform very well on its first day of trading.
- [10:27:00, Medium-term] Anthropic will not build a full-fledged legal application to compete with companies like Harvey.
- [11:31:00, 1-2 years] Traditional marketing automation software will become obsolete because it is not designed for AI agents.
Key Technologies (3)
- Agentic AI / AI Agents: Autonomous AI systems that can perform complex tasks, make decisions, and execute workflows, fundamentally changing how software is used and built.
- Wafer-Scale Chips: Cerebras’s core technology, which involves creating a single, massive chip the size of a silicon wafer to accelerate AI computations, challenging traditional GPU architectures.
- AI Inference: The process of using a trained AI model to make predictions or generate outputs. Cerebras’s story is about providing high-speed inference.
Companies Mentioned (26)
Anthropic · SpaceX · Cerebras · Google · Nvidia · Ramp · AppLovin · HubSpot · Cloudflare · ZoomInfo · Clay · OpenAI · Microsoft · Amazon · xAI · Monday.com · Parker · Gusto · Harvey · Legora · Adobe · SaaStr · Atlassian · Foundation Capital · Micron · SK Hynix
Notable Quotes (8)
If you’re not accelerating, you’re going to be destroyed, right? — Jason Lemkin @ 00:38:00
I’ll never give you 20 times again. I won’t fall for that one this time. — Rory O’Driscoll @ 00:53:00
ZoomInfo’s growth was stolen from it from Clay and friends, and it’s a brutal case study. — Jason Lemkin @ 03:29:00
The enemy of my enemy is my friend. — Rory O’Driscoll @ 04:07:36
Why would I go see Andrew when he was on 20VC and was better? — Jason Lemkin @ 05:02:00
My brain was permanently rewired. I was no longer the same human being. — Jason Lemkin @ 12:10:00
You are all changed. It’s not just a peer group, you’re not the same people. — Jason Lemkin @ 13:35:00
A lot of people think they want to be me, but I promise you when the lights go out at the end of the day, it’s very lonely in my head. — Daniel Dines (quoted by Harry Stebbings) @ 14:13:00
Key Topics
AI Market Dynamics · Venture Capital · Public Markets · SaaS Economics · IPO Market · Founder Psychology · AI Infrastructure · Secondary Markets · Agentic AI
Takeaways
- The AI market is consolidating around compute, forcing major players like Anthropic to secure massive, long-term capacity deals with competitors like Google and even xAI.
- Public markets are unforgiving; SaaS companies showing any growth deceleration without a strong, accelerating AI narrative are being heavily punished, regardless of profitability or beating estimates.
- The concept of ‘Agentic AI’ is a critical paradigm shift. Software categories that cannot adapt to be used by or integrated with AI agents are at risk of becoming obsolete.
- The IPO window is open for highly sought-after AI companies. Cerebras’s 20x oversubscribed IPO indicates massive investor appetite for credible Nvidia alternatives.
- The journey of a successful founder, especially over 5+ years, involves immense personal sacrifice that can permanently ‘rewire’ their brain and psychology, a cost often underestimated.
- Controlling secondary share sales is becoming a standard, critical practice for hot private companies like Anthropic to manage their cap table and prevent market chaos ahead of a potential IPO.
- Even dominant pre-AI companies like ZoomInfo are vulnerable to disruption from newer, AI-native startups like Clay that can commoditize their core data offering.