Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : Value Investing -- Ignore unavailable to you. Want to Upgrade?

To: Spekulatius who wrote (72855)	5/28/2023 10:56:38 PM
From: bruwin	Read Replies (1) \| Respond to of 78748

IMO as this new "AI" (Artificial INTELLIGENCE) phenomenon grows it could, by its very nature, get out of hand and possibly out of the control of those who set it going in the first place, and that may turn out to be BAD for us "Lesser Mortals" who could end up not knowing what's Real or Not Real and Who did What and Why ....

To: Spekulatius who wrote (72855)	5/28/2023 11:13:40 PM
From: Elroy	Read Replies (2) \| Respond to of 78748

with generative AI becoming the primary workload of most of the world's data centers generating information Whoa whoa, wait a minute. Generative AI is some type of software application. A data center moves bits in out of storage super duper fast, does some processing on them perhaps, and moves more bits in and out of storage depending on the outcome of the processing. Data centers don't "do AI" as a primary workload. NVDA and INTC make semiconductors that are logic-based. They don't "do AI". The idea that a $50 million data center is just "doing AI" doesn't make a whole lot of sense to me. I think what is happening is the data center designers are finding that NVDA's graphics processors are much more efficient at handling the applications that run in the data center than are Intel's CPU processors, and the data center architecture is therefore shifting away from Intel and toward NVDA. Great. That's much different from saying the data center has artificial intelligence as a primary workload.

To: Spekulatius who wrote (72855)	5/29/2023 10:56:22 AM
From: E_K_S	Read Replies (1) \| Respond to of 78748

Re: INTC & CPU Servers in Data Center You may be right about the legacy CPU's used in these datacenters. What is the useful life of these CPU's (years ?). Seems like the argument is more about energy efficiency. Qualcom has released their next generation Snapdragon CPUs, very energy efficient but more likely used in cars and mobile (not in data centers) but will support 'at the edge' AI applications. In a recent benchmark test, the Snapdragon 8 Gen 1 chip achieved a CPU performance of 1,000 points, while the NVIDIA A100 chip achieved a CPU performance of 4,500 points. This means that the NVIDIA A100 chip is four times more efficient than the Snapdragon 8 Gen 1 chip in terms of CPU performance. However, it is important to note that the Snapdragon 8 Gen 1 chip is also much less expensive than the NVIDIA A100 chip. The Snapdragon 8 Gen 1 chip costs around $200, while the NVIDIA A100 chip costs around $10,000. INTC also is one of a few companies that own/operate the Fabs. AMD, QCOM, NVDA are ALL fabless. I potentially see a bottleneck in FAB capacity. INTC may/could convert current FAB assets to manufacture specialized chips. They are currently building several new FABs to be online in 3-5 years. They could even license certain chips (AI specific, next generation server chips etc) and manufacture in their facilities and/or sell direct. All these options take time but does not make their assets worthless. Some say many of these specialized chips (ie memory) will eventually become a commodities and have very thin margins. Another factor to consider when new technologies emerge. ------------------------------------------------------ Will the Fabs w/ extra capacity partner w/ the fabless designers? Is/will that be the preferred model? Really depends on the life cycle of the specialized chips. Never see auto's (or EV's) manufacture/design a car w/o owning the manufacturing plant. That too may change especially w/ the next generation specialized chips used in self driving cars. INTC the largest share holder in Mobileye Global Inc Cl A (MBLY). A $ 36 Billion market cap (34% of INTC's current market cap). INTC's market cap only $121 Billion market cap. vs NVDA at a $963 Billion market cap. What will this look like in 3-5 years? I am betting the market cap spread between INTC & NVDA will narrow. Also may/could see more resources focused on specialty high margin applications. This seems to be the path NVDA is pursuing w/ their datacenter & new partners as the total market expands.