AI concentrates chip power
The AI boom is tightening industrial control in a handful of manufacturers and platform companies rather than spreading demand evenly. Taiwan Semiconductor forecasted second‑quarter revenue up to $40.2 billion and signalled a multi‑year AI growth cycle, highlighting how manufacturing capacity is central to the race (benzinga.com). At the same time Nvidia is pushing AI into robotics and has helped revive interest in quantum‑computing stocks, while its CEO describes investing across the whole AI ecosystem rather than backing a single application layer ( ).
The artificial-intelligence boom is concentrating power in the companies that make advanced chips and the platforms that sell the computing around them. (cnbc.com) Taiwan Semiconductor Manufacturing Co., the world’s largest contract chipmaker, said on April 16 that second-quarter revenue should reach $39 billion to $40.2 billion after first-quarter revenue rose 40.6% from a year earlier to $35.49 billion. The company also raised its full-year 2026 revenue growth forecast to “more than 30%” in U.S. dollar terms. (cnbc.com) That demand is coming from the most advanced part of the market. In TSMC’s first quarter of 2025, high-performance computing, the category that includes many artificial-intelligence chips, made up 59% of net revenue, while 7-nanometer-and-below processes accounted for 73% of wafer revenue. (investor.tsmc.com) A chip designer like Nvidia does not run its own leading-edge fabs; it sends designs to manufacturers, which then rely on memory suppliers, packaging specialists and server assemblers. In an April 15 interview, Jensen Huang described that chain as the supply base for a business that could reach “trillion dollars in scale.” (dwarkesh.com) Nvidia is also pushing that stack into new markets instead of staying inside data centers. At its March 2026 GTC conference, the company said its software and hardware now span automotive, healthcare, industrial systems, robotics, telecom and quantum computing. (blogs.nvidia.com) Robotics is one example of how that expansion works. Nvidia’s Isaac GR00T platform is built to train humanoid robots with shared software models and simulation tools, letting the same chip-and-software ecosystem move from cloud servers into factory and warehouse machines. (developer.nvidia.com) Quantum computing is another. Nvidia’s Ising release on April 14 packaged open artificial-intelligence models for quantum-chip calibration and error correction, and quantum-computing shares rallied this week after the launch, with CNBC reporting sharp gains in IonQ, D-Wave Quantum, Quantum Computing and Rigetti Computing. (isingai.net; cnbc.com) Huang said this week that Nvidia tries to invest broadly rather than choose a single application winner. At GTC, he said the conference covered “every single layer” of artificial intelligence, and he pointed to $150 billion of investment into venture startups over the last year. (dwarkesh.com; blogs.nvidia.com) That leaves the industry with a narrow set of choke points: TSMC for advanced manufacturing, Nvidia for the dominant artificial-intelligence computing platform, and a small circle of suppliers around them. TSMC’s April forecast and Nvidia’s March and April product push show that more artificial-intelligence demand is flowing through those same bottlenecks, not around them. (cnbc.com; dwarkesh.com; blogs.nvidia.com)
