Earnings Call, Webinar
The Biggest Bottlenecks For AI: Energy & Cooling
- The opportunity set is characterized as "huge" as tech markets reach unprecedented sizes and companies remain private longer, with the expectation that technology will eventually dominate total market cap.
- AI is projected to serve as a significant tailwind for investments over the next decade, with infrastructure capital expenditure anticipated to reach an annual run rate of roughly $400 billion.
- Input costs for accessing AI models are expected to continue a trend of approximately 100X declines following a 99% drop in the last two years, driven by competition at the model layer and potential capacity scaling.
- Frontier AI capabilities are predicted to improve by a factor of two every seven months, potentially creating market value exceeding $10 trillion, surpassing the mobile and cloud cycles.
- The speaker forecasts that AI will become as ubiquitous as electricity or Wi-Fi, with usage driving efficiencies in US white-collar sectors which represent 20% of GDP.
- Monetization models are expected to reconfigure toward high-end subscriptions priced between $200 to $300 per month and freemium structures, targeting a shift from the current 40 million paying users toward a potential base of 2 billion over five years.
- Consumer engagement is considered a de-risking factor, evidenced by current daily active users spending roughly 28 to 30 minutes daily on products like ChatGPT.
- Public companies may face declining referral traffic and engagement due to AI summaries reducing the need for direct website visits, impacting Fortune 500 consumer engagement strategies.
- Energy is identified as the primary bottleneck for the sector over the next five years, with an anticipated embrace of nuclear power and subsequent innovation in cooling technologies.
- A bifurcation in the AI tool market is expected between emergent usage and tools built for deployment, with stickiness increasing for applications deeply integrated into specific company rules.
- Private markets are expected to remain the primary home for high-growth technology companies for the foreseeable future, though these entities will likely IPO later with increased tender activity to compete for talent.
- Investment activity will focus on "champion" AI apps and American dynamism companies over the next five years, with a strategy of reactive allocation for special research teams and a preference for follow-on investments.
- The firm anticipates significant upside in monetizing the user base for services like OpenAI, expecting pricing power to exceed current levels similar to historical trends seen in social media platforms.
- Gross margins for AI-native applications will be assessed more leniently than mature SaaS, based on the hypothesis that input costs will continue to decline significantly over time.
- A rule of thumb suggests 90% of value goes to end customers while 10% goes to service providers, with the latter still generating massive market capitalization.