Invest Like the Best with Patrick O'Shaughnessy
Dylan Patel - The Infinite Demand for Tokens, Claude Mythos, and Supply Constraints - [Invest Like the Best, EP.469]
· Summarized
Key takeaways
- The demand for advanced AI models and tokens is exploding, with willingness to pay reaching near unlimited levels due to their utility in various high-value applications.
- Businesses that swiftly adopt and leverage AI tools can significantly increase productivity and stay competitive, but risk being commoditized if they do not continuously improve their offerings.
- Supply chain constraints—especially in memory, logic chips, and fabrication equipment—are tightening, which will likely extend lead times and push prices higher, potentially bottlenecking AI infrastructure expansion.
- Rapid progress in AI model capabilities, exemplified by Mythos, is shifting the competitive landscape, favoring those with access to cutting-edge models and compute resources.
- Societal reactions to AI advancements, including protests and backlash, are anticipated and underscore the importance of ethical communication and responsible deployment of AI technology.
Summary
Explosive Growth in AI Spending and Usage
Dylan Patel recounts his firm's rapid increase in AI expenditure, escalating from tens of thousands of dollars to over $7 million within a year. This surge is driven by non-technical teams embracing AI for coding and analysis, significantly reducing the need for hiring and boosting productivity. Case examples include a reverse engineering lab creating GPU-accelerated chip analysis tools and an economist developing an AI-driven economic metric—both projects that formerly would have required large teams. This highlights the broad and fast-growing demand for AI tools and tokens, emphasizing the importance for businesses to harness this technology to stay competitive and avoid commoditization.
Demand for Frontier Models and the Value of AI Tokens
The podcast discusses the unprecedented demand for advanced AI models like Claude Mythos, which exceeds previous capabilities and costs significantly more but offers extraordinary utility. Dylan notes that the willingness to pay for the most capable models is nearly unlimited, fueling continuous innovation and model scaling. The conversation emphasizes that access to the newest, most powerful models—often limited to select enterprise contracts—creates a competitive edge, with high token costs translating into immense value. The rapid progress in model capabilities, cost reductions, and the strategic importance of securing advanced tokens are central themes, as demand continues to outstrip supply.
Supply Chain Bottlenecks and Infrastructure Constraints
Dylan outlines critical bottlenecks across the semiconductor supply chain—memory (DRAM, NAND), logic chips, and fab equipment—that will constrain scaling efforts amid rising demand. He highlights how lead times for capacity expansion are extremely long, often extending into 2028, causing prices for memory and other components to soar. The hardware industry is experiencing record margins and prepayments, with suppliers like TSMC and ASML operating at full capacity and investing heavily in expansion. These supply constraints threaten to bottleneck AI infrastructure development, making chip, equipment, and raw material shortages a major concern for future growth.
AI Model Progress, Supply Dynamics, and Competitive Landscape
The episode explores rapid advancements in AI models—particularly Anthropic's Mythos, which has leapfrogged previous generations and is very cost-efficient. Dylan notes that AI labs are competing to release the most capable models, with Anthropic leading due to strategic compute management. The discussion touches on the role of hardware innovation, intensive CapEx, and the increasing difficulty for other players to keep up. Additionally, the scarcity and high costs of tokens are central to the economics of AI, with demand far exceeding supply, driving margins and potentially leading to competitive advantages for those with early access and strong infrastructure.
Future Outlook: Societal and Economic Impacts of AI
Dylan predicts significant societal reactions to AI advancements, including large protests against AI’s influence and potential misuse. He emphasizes that the accelerated progress in models, coupled with increasing resource concentration, could lead to societal disruptions and increased inequality. The conversation underscores the importance of responsible communication by AI companies and the need to focus on practical, present benefits rather than speculative future capabilities. Strategic access to the latest AI models remains a key factor for economic success, with a warning that without proactive management, societal backlash could intensify.
Sponsors
- Ramp · Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth. Learn more at ramp.com slash invest.
- Vanta · Vanta automates security and compliance for over 16,000 fast-moving companies, helping them stay audit-ready and secure across their environments. Get $1,000 off at vanta.com slash invest.
- Richline · Richline is the first end-to-end system of record with embedded AI for investment management, offering portfolio accounting, reconciliation, reporting, trading, and compliance on a unified platform. Request a demo at ridgeline.ai.
- Colossus · Visit colossus.com for in-depth profiles of influential founders, investors, and companies, plus access to the quarterly publication and episodes.
- Rogo · Rogo is an AI platform tailored for finance, building agents that understand how Wall Street works, from diligence to modeling to analysis. Learn more at rogo.ai/invest.
- Ridgeline · Ridgeline has built a real-time, modern operating system for investment managers, covering trading, compliance, reporting, and more. Discover what they can do for you at ridgelineapps.com.
Show notes
This is my second conversation with Dylan Patel. Dylan is the founder and CEO of SemiAnalysis, where he tracks the semiconductor supply chain and AI infrastructure buildout.
This conversation is about the supply and demand of tokens. On demand, Dylan describes something completely explosive. He explains why the frontier model is the only model anyone wants, and willingness to pay for it is nearly unbounded. His own firm has gone from tens of thousands of dollars in AI spend last year to seven million this year.
On supply, we walk through the bottlenecks across memory, logic, and fab equipment that will determine how fast any of this can scale.
We also cover Claude Mythos and what the leading labs need to do to fix their growing public perception problem.
For the full show notes, transcript, and links to mentioned content, check out the episode page here.
-----
Become a Colossus member to get our quarterly print magazine and private audio experience, including exclusive profiles and early access to select episodes. Subscribe at colossus.com/subscribe.
-----
Ramp’s mission is to help companies manage their spend in a way that reduces expenses and frees up time for teams to work on more valuable projects. Go to ramp.com/invest to sign up for free and get a $250 welcome bonus.
-----
Trusted by thousands of businesses, Vanta continuously monitors your security posture and streamlines audits so you can win enterprise deals and build customer trust without the traditional overhead. Visit vanta.com/invest.
-----
WorkOS is the infrastructure B2B and AI-native companies use to sell to enterprise. It covers everything enterprise security requires: SSO, SCIM, RBAC, Audit Logs, AI governance, and more. Trusted by 2,000+ fast-growing companies, including OpenAI, Anthropic, Cursor, and Vercel.
-----
Rogo is the AI platform for finance. They're building agents for Wall Street that are trained to understand how bankers and investors actually do work: from diligence and modeling, to turning analysis into deliverables. To learn more, visit rogo.ai/invest.
-----
Ridgeline has built a complete, real-time, modern operating system for investment managers. It handles trading, portfolio management, compliance, customer reporting, and much more through an all-in-one real-time cloud platform. Visit ridgelineapps.com.
-----
Editing and post-production work for this episode was provided by The Podcast Consultant (https://thepodcastconsultant.com).
Timestamps:
(00:00:00) Welcome to Invest Like The Best
(00:02:29) Intro: Dylan Patel
(00:03:09) Semi Analysis AI Spend: Zero to $7M
(00:05:16) Real-World Examples of Claude Code
(00:11:41) Token Demand: “Completely Explosive”
(00:14:48) Why Everyone Wants the Frontier Model
(00:15:36) Mythos: Biggest Model Capability Jump in Two Years
(00:20:54) Fear of Rapid Model Progress
(00:23:45) Robotics as the Next Demand Wave
(00:26:03) Scaling Laws & Compute Efficiency
(00:27:24) OpenAI vs. Anthropic
(00:31:33) Supply Side: Bottlenecks Across the Stack
(00:33:26) TSMC CapEx Could Cause a Shortage
(00:36:45) CPUs, ASICs, and FPGAs
(00:40:12) Tokenomics
(00:42:20) Protests & AI Backlash
Follow this show on AudioBrevity and new episodes get summarized for you. Create a free account