EXCLUSIVE: Nvidia Keeps AI Profits, TSMC May Want a Bigger Cut

Nvidia Corp (NASDAQ:NVDA) has built one of the most profitable businesses in the AI boom without owning the fabs that manufacture its chips. But as demand for AI accelerators keeps pushing foundries to expand capacity, Josh Kaplan — who manages the MVSMH index that underlies VanEck Semiconductor ETF (NASDAQ:SMH)— sees a possible shift in where the semiconductor industry’s economic profits land. Taiwan Semiconductor Manufacturing Company Ltd. (NYSE:TSM) could have leverage, according to Kaplan. • NVIDIA stock is trading near recent highs. What should traders watch with NVDA? Nvidia’s Margins Create a Target Nvidia’s advantage in AI has largely come from its ability to combine powerful chips, software and networking into a platform customers are willing to pay up for. The company reported a 71.1% gross margin for fiscal 2026, underscoring just how much economics have accumulated at the chip-design layer. But Nvidia does not manufacture those advanced chips itself. It relies on foundries such as TSMC to turn designs into finished silicon. Kaplan, who is head of Research & Investment Strategy at MarketVector, says that dependence matters because foundries control the capacity needed to translate AI chip demand into actual shipments. "Nvidia and other chip designers rely completely on allocation of capacity from foundries," Kaplan told Benzinga in an exclusive email interview. That creates a potentially important imbalance: Nvidia can capture enormous profits from AI demand, while leading foundries control one of the most difficult parts of the supply chain to replicate. Read Also: Nvidia's AI Boom Hits Its First Margin Wall: Memory Prices TSMC Has More Leverage Than It Looks Kaplan specifically pointed to TSMC and Intel Corp (NASDAQ:INTC) as companies that can "single-handedly enable chip designers to increase volumes." That leverage becomes more significant as AI chips become increasingly complex and demand for advanced manufacturing and packaging grows. Recent industry research has highlighted the same dynamic. JPMorgan Asset Management noted that Nvidia, Broadcom Inc. (NASDAQ:AVGO) and Advanced Micro Devices, Inc. (NASDAQ:AMD) collectively spend a significant portion of their cost of goods sold with TSMC, while describing advanced semiconductor manufacturing as highly concentrated. For Nvidia, that means manufacturing is not simply a cost line. It is a critical dependency. And Kaplan sees a possibility that the economics eventually reflect that. The AI Profit Pool Could Shift "If I’m Taiwan Semiconductor and I’m watching Nvidia’s gross margins grow and grow, I’m thinking ‘hey I want a piece of that,’" Kaplan said. He added that he "wouldn’t be surprised to see some of the economic profits in the semiconductor industry shift a bit from designers to manufacturers." That does not mean Nvidia suddenly loses its pricing power or that TSMC captures Nvidia-like margins. It points instead to a gradual redistribution of value as manufacturing becomes an increasingly strategic bottleneck in AI. For investors, the important signal may therefore be less about Nvidia versus TSMC and more about where the AI semiconductor profit pool is accumulating. Why it Matters Nvidia’s margins remain a defining feature of the AI trade, but Kaplan’s thesis puts the foundry side of the equation under a different spotlight. As AI chip volumes rise, investors may want to watch whether more of the industry’s economics begin flowing toward the companies that control advanced manufacturing capacity. Read Also: Nvidia Built the AI Brain — Now TSMC Wants to Give It Eyes Image via Shutterstock