Filtered to AI Hardware & Semiconductors · show all articles
Applied Materials builds layers up. Lam Research carves them back down — the etch step that turns a deposited film into an actual working circuit pattern, and its own genuinely hard engineering problem.
Deposition builds the layer. Etch cuts the pattern. Neither step tells you whether it worked. That's KLA's job — inspection and metrology, the unglamorous layer that decides what actually ships.
Semiconductor stocks are more concentrated in the S&P 500 than tech was at the dot-com peak, and also backed by real demand this cohort has documented elsewhere. This briefing weighs both without picking a side.
Panther Lake is shipping on 18A, Intel's most advanced node in over a decade. What the company has not yet done, as of this briefing, is prove the node can win business it didn't already have.
18A proved Intel could build a competitive process again. 14A is the node where the industry finds out whether anyone outside Intel wants to build on it — and the window to find out is now open.
A 23.2% share of the global automotive semiconductor market, and slower EV adoption than expected — while AI-driven power demand quietly becomes the segment actually lifting the company's 2026 guidance.
A next-generation product usually promises more performance than the last. Huawei's 2026 Ascend roadmap breaks that pattern on paper — and the reason why is a rational, deliberate trade-off, not a stumble.
By 1986, Japan held roughly 75% of the global DRAM market. By the late 1990s, that lead belonged to Korea, and by the 2000s, Taiwan's foundry model had reshaped the whole industry around it. This briefing covers why, not just when.
Groq built its whole identity on the claim that GPUs are the wrong chip for AI inference. Then Nvidia got involved. Here's the short version.
In 2018, GlobalFoundries walked away from the race to the smallest transistor entirely. In 2026, that decision looks less like a retreat and more like a specific, working strategy.