Sourcery: Why Coatue Follows the Gigawatt — AI, Semiconductors & Market Shifts

**SPEAKER_1** (0:01)
When private AI companies are scaling faster than almost anyone thought possible, the old rules of tech investing start to look obsolete. Today we're diving into a fascinating conversation between Molly O'Shea and Jaimin Rangwalla, CIO of Coatue's public markets business. They cover everything from AI and semiconductors to market structure and what it means to invest through a once-in-a-generation shift.
And it starts in Coatue's new office, which is really interesting. Jaimin describes this bullpen layout where AI researchers, data scientists, and analysts all sit together. He says "We all came in Monday morning, not knowing what to expect," but the design is intentional—it's built for fast collaboration. There's even an office soccer group that's become part of the culture story.
Right, that intensity seems core to how they operate. But what really stands out is why Jaimin thinks the AI opportunity feels different. He says the important thing isn't just size, but speed. Adoption curves, revenue curves, user curves—they're all rising faster than in prior technology cycles.
OpenAI reaching nearly a billion users is a perfect example of how steep those curves have become. And Jaimin's view is that the market is still early because the curve hasn't flattened yet. That's a bold take given how much hype already exists around AI.
It shapes how Coatue maps the entire AI ecosystem. Here's what's fascinating—Jaimin says they've moved from following the GPU to following the gigawatt. Power has become the atomic unit of growth. Instead of old sector buckets, they're tracking inputs to power, buyers of power, and users of the tokens that power produces. That's a fundamental reframing. And one of his biggest surprises is how persistent the shortages have been. He says, "I am surprised at how the tightness has persisted for this long."
Memory, power, optical components, labor, data center capacity—all constrained at once. Which makes AI harder to model but also signals the opportunity is much bigger than many realize. Now, Jaimin's career gives him perspective here. He started in semiconductors in 2007, right when the iPhone arrived. He remembers sending staff to stand in line for one. That taught him innovation can reshape industries quickly and what looks cyclical can become structurally powerful. His path to Coatue went through NYU, two years at Merrill Lynch in investment banking, then into public markets investing. The lesson he learned early was simple: great careers come from finding a few big ideas and letting them compound. At Coatue, he describes it as "sprinting a marathon."
That phrase captures it perfectly. But let's talk scale, because this is remarkable. The biggest Mag 7 IPO was Meta at about $100 billion.
Now OpenAI, Anthropic, and SpaceX are valued in the hundreds of billions, with SpaceX above a trillion. Jaimin says, "The fact that you now have private companies breaking into pretty much the top 25 in the world before they even go public, it's an unprecedented time." And the revenue growth matches that scale. Anthropic and OpenAI are adding billions in revenue per week sometimes— more than many SaaS companies generate in a year. That confirms AI isn't a niche trend, but a market broad enough to support enormous enterprise value. The conversation gets technical around tokens and agents. Jaimin defines a token as "the unit of thought by any AI model." Agents have become a huge unlock because they can now launch other agents—less human supervision, more depth of work, more parallel execution. And that increases demand for semiconductors and memory dramatically. More agents mean more CPU work, GPU work, and far more memory footprint. Jaimin imagines a digital population explosion where each person commands thousands of virtual workers. That creates more hardware demand than most people appreciate.
On the market side, he draws a sharp line between sellers of shortage and buyers of shortage. Sellers like memory, optical, power, and infrastructure companies benefit from scarcity and pricing power.
Buyers, including hyperscalers with heavy capex, see multiples compress. Though Google and Amazon are interesting hybrids. Optical components were a big surprise— moving from oversupply to critical infrastructure overnight. And Molly pushes on sentiment because headlines sound deeply negative. Jaimin's blunt: "Fundamentals matter more than sentiment." Earnings growth remains strong, multiples haven't exploded, companies are beating estimates. The episode closes with Jaimin praising Philippe Laffont as both a great investor and builder. He says Coatue is entering another inflection point, and the mindset is constant curiosity about what's changing and how to capitalize on it. For AI investing, that's the whole game.