What's worrying me if how they seem to have decoupled multiple steps in the process to achieve equilibrium in phases. This thing could work and get high yields, since Huawei could use their own process to make their own products and complete multiple learning cycles. I am sure the Chinese government could support them if they are having negative cash flow in the fab for a prolonged period.

---

Intel claims to be able and willing to fab both x86 and ARM for IFS customers. Also it's being priced in as true because Ming-Chi Kuo is a reputable leaker of apple stuffs.

---

The biggest "quiet part" with all this AI talk. No fab has absolutely boomed on it. Like one tweet away from INTC doubling

---

Don't worry guys. Enron Musk will come save the chip industry. Because they're a chip fab this week. Or was it robotaxi? I can't remember.

---

What I'm saying is that Google & Broadcom should be able to compete with the big boys for datacenter marketshare. The chips can theoretically be used for more MUCH than data science and training LLMs. Two big question marks though: 1. Are Google & Broadcom willing and able to secure the requisite fab capacity? 2. Is Google willing to dedicate resources towards developing translation layers & other tools to help developers migrate software that wasn't written with TensorFlow?

---