NVDA
TSM
The chip party keeps getting bigger
Taiwan Semiconductor Manufacturing Co. basically looked at the AI craze and said, “Yeah, but what if this is still early?” In presentation materials ahead of its technology symposium in Hsinchu, the company forecast the global semiconductor market will swell past $1.5 trillion by 2030, up from its prior $1 trillion call.
That’s a pretty massive upgrade. And it’s not being driven by phones or cars doing their best impression of smart devices — the heavy lifting is coming from artificial intelligence and high-performance computing, which TSM says could make up 55% of the market.
AI is still hogging the bandwidth
The company also said AI accelerator wafer demand could jump 11-fold between 2022 and 2026. Translation: the infrastructure train is still leaving the station, and everybody wants a seat.
That matters because TSM sits at the center of the semiconductor universe. If Nvidia needs more advanced packaging, if AI chips need more wafers, if the whole ecosystem keeps scaling — TSM is the tollbooth.
Building like the demand’s not slowing down
TSM isn’t just talking a big game; it’s spending like it believes it. The board approved $31.28 billion in capital spending to expand advanced chip manufacturing, build new fabs, and install the systems needed for AI, 5G and high-performance computing demand.
A few other nuggets from the update:
- Output for its 2-nanometer and next-gen A16 technologies is expected to grow at a 70% CAGR from 2026 through 2028.
- CoWoS advanced packaging capacity is projected to rise more than 80% on a CAGR basis from 2022 through 2027.
- In Arizona, the first plant is already producing, while more capacity and advanced packaging are on the way.
Why investors should care
This is TSM telling the market the AI buildout is still in its early innings, with real dollars attached. That’s bullish for the chip supply chain — especially names like Nvidia that depend on TSM’s manufacturing and packaging muscle — but it also raises the bar for execution. Big forecasts are nice; actually building the plants without tripping over your own supply chain is the part that counts.
Big picture: the AI boom is no longer just about who sells the chips. It’s about who can make enough of them fast enough to keep the whole circus moving.