Why the Real AI Bottleneck Isn't Compute | Howard Chan, Kurv Investment Management
Howard Chan is back on Behind the Ticker for his second appearance, celebrating almost exactly the two-year anniversary of KQQQ — Kurv Investment Management's first ETF. When Howard first sat down with Brad two years ago, the conversation was about theory. This time he brings two years of live performance data validating the approach and a new fund tackling one of the most under-appreciated bottlenecks in the AI trade. KMEM — the Kurv Memory Select ETF — launched on July 1st with a specific thesis: while the market has spent years pricing NVIDIA and the compute layer of AI, the next binding constraint is memory. Ninety percent of the world's memory chips are made by three companies — Micron, Samsung, and SK Hynix — and all three are sold out through 2028. Howard walks through why the shift from commoditized memory to specialized HBM chips has created a moat where none existed before, why SK Hynix just reported 250% revenue growth and 500% profit growth, and why the physical constraints on expansion (fab costs, ASML equipment waitlists, workforce buildout) mean this is a two-to-four-year story rather than a quarter-to-quarter one. He also explains why the secondary effect — memory costs pushing up Apple, Xbox, and MacBook prices — signals how deeply this constraint is filtering into the broader economy. Howard also covers the two-year update on KQQQ, which has kept pace with and in fact outperformed its mega-cap tech underlying while generating 15-18% distribution yield through selective options overlays — writing calls only on positions with limited upside rather than mechanically across the portfolio. He walks through why the pairing of KQQQ with a dividend equity ETF gives an advisor a more complete sector picture than either alone, and hints at the tax-efficient portable alpha strategies coming from Kurv in the months ahead.





