The Long Term Investor

AMD Confirms EPYC "Venice" Zen 6 CPUs Launch on July 22-23
https://www.techpowerup.com/350648/...us-launch-on-july-22-23-at-advancing-ai-event

Sexy! Too bad I am not filthy rich and don't really have a use for this much raw power.

Now for the elephant in the room:

To fully max out a 6th-Generation AMD EPYC "Venice" CPU to its absolute limit, it would cost between $110,000 and $240,000 for the RAM alone. [1, 2, 3]
Because the Venice platform utilizes a massively upgraded architectural layout, achieving its full capacity requires cutting-edge, ultra-high-density enterprise memory modules. [1, 2]
The Technical Requirements

• The Memory Ceiling: An EPYC Venice processor on the SP7 socket supports an unprecedented 6 TB of system RAM per socket. [1]
• 16-Channel Architecture: Venice expands the memory subsystem to 16 distinct channels. To maximize performance and capacity at 1 DIMM Per Channel (1DPC), you must use exactly 16 memory sticks. [1, 2, 3]
• The Sticks Needed: To hit 6 TB across 16 channels, you have to use massive 384 GB or 256 GB modules (e.g., highly specialized 3DS RDIMMs, MRDIMMs, or MCRDIMMs). [1, 2]

The Estimated Cost Breakdown
Enterprise pricing fluctuates drastically based on bulk procurement and vendor discounts, but current market rates for these bleeding-edge high-capacity sticks dictate the following tiers:
Configuration Setup Module Type Est. Cost Per Stick Total RAM Cost (16 Sticks)
Max Capacity (6 TB) 384GB DDR5 3DS RDIMM / MCRDIMM $12,000 – $15,000 $192,000 – $240,000
Performance Max (4 TB) 256GB DDR5 8800MT/s MRDIMM / MCRDIMM $7,000 – $11,000 $112,000 – $176,000
Why is it so expensive?

1. MRDIMM / MCRDIMM Premium: Venice hits its record-breaking 1.6 TB/s of bandwidth by leveraging these next-gen multiplexed modules. They effectively double the data rate internally over standard RDIMMs, commanding a steep price premium. [1, 2, 3, 4]
2. Extreme Density: 256GB and 384GB sticks require complex 3D TSV (Through-Silicon Via) stacking of memory dies, which suffers from lower manufacturing yields compared to mainstream 32GB or 64GB server RAM. [1]

 

A bit off topic, but I had the occasion to head to the polls recently for primary voting and a couple of local issues. As usual, I did a bit of research on the candidates to learn more.

It was like sorting through trash to be frank about it. There was nothing really worthy about any of them. I remember thinking..."Is this the best we have to offer or choose from?" I felt like the ballot would best be served by a shredder, but there were a couple of local issues I wanted a voice on. So, I continued with my civic duty and cast my vote.

It seems we have equally bad choices on either side and offer up the worst options either can come up with.

 

Another week in the books for the market.

Again, I missed most of the week and associated noise. Looks like most of it is just a repeat of the same old things anyhow. The broad market notched another green week it appears.

 

I don't post much media related articles anymore for a number of reasons. I wonder how much of this type of thing is going on in the so called "AI race" among different companies.

Apple sues OpenAI alleging trade secret theft

 

"The system is not broken. It's working as intended."

Obligatory social critique by the GOAT:


The sociopathic elites will always rise to the top. One cannot beat the system. One can only observe, learn, and grab a slice of the pie while they can.

 

I think it is rampant. I am not a fan of Apple, but I hope they spank that little mouse Sam Altman.

 

Congrats to Lori’s team. England wins their match against Norway.

 

Well, it's that time again. Earnings reporting. Seems we are barely done before it begins once again.

Of course, the big banks/firms will kick it all off next week. It will then pick up and continue with more and more notable companies reporting.

We continue to roll along through this year.

 

Let's see if I can further lower my social status on Stockaholics by sharing my view of equity behaviors. This ought to go well!

I bought my first stock in 1983 after following the market for a few years. I went all in with my $2400 that was saved from multiple summer jobs, part time work, and living like a vagrant. The markets had few gains since the 1970s. People were distrustful of stock and weary of another crash. Oil was almost worthless. The economy was in recession.

At the time, there were recognizable stocks with 2:1 PE ratios and 4:1 was common. There was the odd Cinderella company with 15:1 ratio. These were pretty much exclusively penny stocks and companies that carried ridiculous levels of risk like mining and junior oil. The generally accepted PE upper limit in my investing group was 7:1.

In 83, it seemed like the market was extremely stable. Stocks would go multiple days in a row with identical closing prices. Previous day closes were provided by the Morning Line Report in the morning paper and that was plenty timely.

20 years on and several crashes later, the market was a whole lot more frothy. Retail investors were pouring into the market and I estimate they were already causing ticks and noticeable fluctuations but big institutions and value still dominated stock prices.

By 03, PE 7 companies still existed but it took a lot of searching. We didn't have the myriad of charting and research web sites we have now. We still didn't even have anything remotely close to live quotes.

20 more years and the markets have gone mental. Sometime in the mid to late 2000s, people started moving 100% of their 401K into VOO. That transition was mostly complete by about 2013. You can see the inflection point and higher slope that starts in 2009.

When is the last time you saw a blue chip at PE 7? When is the last time you saw a blue chip at PE 15? What even is a blue chip, anymore? How many stocks have a PE that is not infinite? (ie: they are profitable) ... maybe 50%? Back in the day, an infinite PE would cause a stock to plummet. People worried about bankruptcy.

When is the last time an IPO failed? They used to fail all the time. Perhaps 60% succeeded in the 70s and 80s.

The key, IMO, is the 2009 S&P 500 slope inflection. That's when gains became significantly higher. Money was pouring in. Prices were started to soar. Merely being included in the S&P 500 index meant a huge windfall for investors.

Is this slope sustainable? Does the S&P 500 have to gain forever? Is an S&P 500 investment safe? What would happen if the S&P 500 fell significantly and people switched their 401K to FI or more conservative investments?

These are the questions. As long as there is confidence, the S&P 500 should do very well. It's not just about absolute confidence. It's going to come down to more secure alternatives. It's hard to imagine China, Europe, India, or any other country being as secure as the S&P 500 but that doesn't mean it will always be the best choice. Maybe Singapore?

The idea that everything will continue as it has forever is a possibility but not a certainty. Personally, I believe we are not at but not far from a historic inflection point.

 

Some interesting points as usual.

A few thoughts from my point of view. I think, in general, folks are always searching for a better mousetrap. Sometimes we focus too much on the trap instead of the overall plan put into catching the mouse.

I always find the Callan Periodic Table of Investment Returns interesting to view. I think this one goes from about 2006-2025. There are others that go further back.

As investors, recency bias is a real thing. It can get us into trouble both ways. Sometimes the easiest thing to do is the one that requires us to do the least.

https://www.callan.com/periodic-table/

 

You bring up the point of VOO !!! And 401K's / retirement accounts , as long as there is a steady stream of fresh money flowing into VOO and VOOG and SPY, the market is going to keep going up. Simple economics, more money chasing a set number of shares of stocks, share prices go up. The moment that chain gets broken, and people quit feeding the market with money from their 401K's and other retirement accounts , the jig will be up.