E396: The Future of Compute, Data Centers, and AI

Episode Summary

Executive Summary: Kush, founder and CEO of Orin, argues that compute is becoming the economy’s core strategic commodity, analogous to oil in the past, and needs liquid markets for price discovery and hedging. The discussion covers compute volatility, inference-provider demand, AI’s infrastructure stack, data center buildout, energy constraints, geopolitics, and why Orin’s transaction-based index targets the model and data center layers.

Main Topics: Compute as the next strategic commodity (Priority: 5/5): Kush frames compute as a national strategic asset, comparing it to oil and arguing that modern economies and AI systems increasingly depend on it as a core input. Why Orin created a compute exchange (Priority: 5/5): Orin is building a liquid market for compute so enterprises and providers can hedge price risk, improve discovery, and transact around real supply-demand conditions. Volatility, demand spikes, and customer fit (Priority: 4/5): The company’s early users are mainly inference providers and other compute buyers/sellers facing spiky demand and uncertain supply, which makes short-term hedging useful. AI ecosystem stack and market positioning (Priority: 4/5): Kush maps AI into application, model, infrastructure, data center construction, energy, and chip layers, and says Orin sits between model companies and data centers where hedging is most needed. Data center buildout, energy, and capital markets (Priority: 4/5): The conversation emphasizes massive capital flowing into data centers, with energy and land becoming the key bottlenecks for future compute capacity expansion. Competition, geopolitics, and the AI arms race (Priority: 3/5): He argues the U.S. must build compute capacity faster than China, viewing data centers as both economically and strategically important in national competition. Startup execution and AI-native hiring (Priority: 3/5): Kush describes building a hypergrowth startup with a young team, fast customer support, and hiring developers who already use AI tools effectively.

Key Arguments: Compute will become the largest economic input for enterprises, similar to how oil powered the industrial economy. A liquid compute market is needed because enterprises and providers face volatility and need hedges, futures-like instruments, and price discovery. Inference providers have particularly spiky demand, so short-term compute procurement is more appropriate than long-term fixed arrangements. Compute volatility is driven by supply-demand imbalances, new model launches, and fresh data center capacity coming online. The market should be based on transaction data, not indicative offers, to ensure the index reflects real ground truth pricing. Data centers and compute infrastructure are strategically important assets, and disruptions to them can slow national productivity. The U.S. must accelerate energy and data center buildout to stay competitive with China in AI. OpenAI has gained an advantage by securing more compute capacity, while Anthropic has historically been more constrained. The current compute boom may contain bubble-like behavior, but underlying demand is so strong that it may take years to appear. Startups can compete through speed, quality, and customer responsiveness, especially by hiring AI-native talent.

Data Points: Projected compute market size: $7 trillion - Kush cites this as the projected size when sizing the compute market. Oren founding timeline: 9 months - He says the company started in September and is now about nine months old. Founder age: 22 - Kush says he is 22 years old. Data center buildout in the near term: 3 to 5 gigawatts - He estimates that amount of compute/data center capacity is coming online in the next couple of years. Historical capacity growth: Double every year - He says the amount of compute coming online has been growing roughly 2x annually. Customer response time: 5 minutes - Oren emphasizes rapid customer support as a differentiator. Traditional consulting response rule: 15 minutes - Kush references this as the standard he adapted for customer service discipline. Old GPU/server pricing: H100 server for 180K - He says used H100 servers can still cost around this level, showing strong demand. New GPU/server pricing: 220K - He says new H100 servers go for about this much. Depreciation schedule: 5 years - He references that chips are often depreciated over five years, yet their market value is rising. AlphaSense expert call coverage: 240,000+ transcripts - Mentioned in the ad read describing AlphaSense’s expert call library. AlphaSense customer base: 75% of the world’s top hedge funds - Mentioned in the ad read describing platform trust and usage.

Pivotal Quotes: "I think the same is true for compute going forward. It's a national strategic commodity that needs to be treated that way and needs to have a market built around it." — Kush: Explaining why compute should be treated like oil or other critical commodities. "We basically create an exchange for compute. The reason is we think compute is going to become the next commodity of the future." — Kush: Describing Orin’s core business and its rationale. "At the end of the day, it's whoever has the most resources, in my opinion, will be able to win this race." — Kush: Discussing the AI competition among major model companies and the role of compute capacity.

Implications: Compute is emerging as a strategic, tradable input with geopolitical and economic consequences. Expect more data center investment, tighter energy planning, and greater demand for market structures that hedge compute risk and improve allocation.