The Price Of Memory

 The title may sound poetic but the subject is prosaic :)

I was curious at the large earnings upside surprise announced by Micron Technologies a few days ago.  The company makes commonplace DRAM chips and other such storage products used in just about everything these days.  It’s basically an electronics commodity company, albeit sophisticated.

The price of memory has been coming down steadily for decades following Moore’s Law, see chart below.

With prices always dropping, the industry’s profitability was always under pressure. Stock prices suffered for years as companies tried to make up in volume what they couldn’t achieve in profit margins.

Until AI and data centers came along and created a large extra demand for common “low tech” commodity semiconductors. Prices doubled, even tripled, within just a year or two.  Since fabrication costs stayed the same, more or less, profits soared.

Share prices for the likes of Micron, Seagate and Intel exploded on the upside with gains of 200-300% in just a few months. Will it last? 

Interest Rates Disconnect

Despite recent Fed rate cuts, long Treasury bond yields remain stubbornly high.  The chart below shows 3 month Tbill rates (black line) and 30 year Tbond yields (red line).  The recent disconnect is obvious, though not uncommon in and by itself, as the chart also shows. However, the 2008-15 period was one of ultra low short rates from the Fed's QE to get us out of the GFC.

More recently, short and long rates fell and rose in tandem during the COVID scare.

So, what is happening today? Why the disconnect?

IMHO, two reasons:

1. The market does not really believe that inflation is beat.  President Trump's constantly pressuring the Fed for lower rates does not help, either.

2. Credit concerns about the US.  Ray Dalio, in particular, has been very vocal on the subject.

While everyone is focusing on the stock market and its love affair with AI, this is what AI has to say when asked about Treasury bonds (ChatGPT):

Q - "Are there concerns about the creditworthiness of the US government?"

A - "Yes — there are significant concerns right now about the credit quality (creditworthiness) of the U.S. government, and these concerns are being reflected in how major financial institutions assess U.S. government debt. The gist is that, while U.S. Treasury securities remain one of the most widely held and liquid assets in global markets, there’s a growing debate among investors, policymakers, and rating agencies about how sustainable U.S. public finances are over the long term."

The answer then continues with more specific points and concludes with:

"Bottom line: U.S. government debt is still widely viewed as a relatively safe investment compared with many others, but recent downgrades and ongoing fiscal pressures have raised legitimate concerns about its credit quality among ratings agencies, investors, and economists."

Here's my bottom line: The US federal government has to get its fiscal act together.  Raise taxes, cut spending, stop promising kids' accounts and bonuses, stop announcing defense spending in the hundreds of billions.  But, more than anything... raise taxes.  I know this is anathema to most Americans, but what is the alternative?  After all, other than defense, almost all other spending goes to holy grails like Social Security, Madicare/Medicaid.

Here's some ideas:

1. Fuel tax - bring it in line with Europe, at least partially.

2. Wealth taxes - they won't raise huge amounts, but at least it shows a commitment to take actions.

3. National VAT/consumption tax.  That's the one that will raise the most money, by far.

The End Of The Yen Carry Trade

The news are everywhere, particularly amongst the breathless LinkedIn and Facebook pundits: The recent pivot of BOJ towards higher rates marks the end of the yen carry trade, so they predict all manner of imminent market disasters. Well..

Given that BOJ has been telegraphing its intentions for months, markets had -  of course - already pivoted away from the yen as a low cost funding currency.  Just take a look at the data from the Bank of International Settlements: Yen credit growth to non-residents (aka the carry trade, mostly) plunged to zero many months ago (panel C below).  Therefore, don't expect a sudden plunge in markets just because the yen carry trade is no longer operable - participants exited long ago.

What has taken its place? The euro carry trade (panel B), since euro interest rates are lower than dollar rates, though probably not for much longer if the Fed keeps cutting rates.

Note: the BIS data are current up to 2Q2025.  Data for 3Q2025 will be available early 2026.

Quantum Computing Questions

I readily profess that my working knowledge of computing goes back to FORTRAN IV - yes, punchcards and all! Therefore, today I use computers as a black box: I input stuff and I get back stuff, and I know nothing of what happens in between, or how it is done.

So, I have a question: will quantum computing eventually make "regular" computing obsolete?  I asked AI and it said no, it won't and that I should think of quantum computing as a "truck" used for carrying very heavy loads and regular computing as a "car" used to carry the week's groceries.

But I wonder... back in my own ancient days a computer center was a huge airconditioned room with dozens of disk drives.  Today, its computing power and capabilities fit in my phone! If back then I had asked the computer center manager the equivalent question, I am sure he would have given me the same answer as AI: use your HP calculator for your daily needs and come to us for the "heavy" stuff.

So, dear reader, will quantum computing become tomorrow's standard? Will qubits kill bits or will they live happily ever after side by side? And if so, what does this mean for today's GPU darlings? 

If any of you have answers I will be very grateful to hear them! Thank you in advance!

H.

The China Syndrome

 Almost from the start of this blog what seems eons ago I pointed out that sending manufacturing to China in the name of consumer affordability and corporate profitability was a serious strategic mistake.  Events have proven me right.  Just look at what is happening in the auto industry in the US and Europe.

What industry is next? Semiconductors and AI.  Reuters just reported that China is rapidly achieving its own AI chip technology, almost a decade ahead of projections. Given the current drunken craze in this sector, the news ring a bell of sobriety.  China is determined not to be left behind in anything, let alone AI.

Can the West compete? Yes, of course. But not in a face-to face bruising battle.  The results of the Trump tariff war prove it - China used its muscle and won, no way to hide it.

So, play smart. Give up the consumer spending model and focus on quality, not quantity.  

Payrolls In The US

 The Bureau of Labor Statistics announced its loooong awaited non-farm payroll numbers today, delayed due to the government shutdown.  Here is the most relevant chart:

My observation: looking at the 3- month numbers (most relevant because of the long shutdown), it's all about (a) the loss of government jobs and (b) the gain in just one category, private education and health services.  Bottom line: jobs are significantly stronger than the headline number where it really matters, the private sector.

The Fed will be looking very carefully at this, I'm sure.

High Prices For US Stocks

Something has been bugging me for several weeks now: prices for US listed stocks are very high.  I mean absolute prices, not valuations based on P/E, P/B, etc. For example, yesterday's price of a single Tesla share was $475, likewise for Microsoft, also at $475, Meta was $647, etc.  Doing a bit of quick research, the median price for S&P 500 stocks recently was approx. $120 and for NASDAQ 100 it was $245.

By comparison, a study by the Fed showed that in 2000 the equivalent median prices were $29 and $25.

Does this mean anything practically? Yes, and no.  No, because it is the total market cap that really matters.  And yes, because the higher they fly the longer the way down.  Meaning, the absolute losses for a shareholder are now potentially significantly bigger.

Was there a similar past occurrence in US markets? Yes, in 1929: for example, the price of RCA (aka "radio") was around $550 in September 1929.  It ended up at $15 in 1932.

I don't mean anything by the above, it's just a weird observation.

Bank Required Reserves For US Banks? Zero

 A kind reader (thanks AKOC) asked what are the required reserves for US banks - ie deposits at the Fed or currency in their vaults.  As of 2020 the answer is..zero.

Here is a helpful chart.

So, why zero?  Because except for macroprudential reasons, holding reserves doesn't truly provide any benefit.  Instead, banks must conform to a Liquidity Coverage Ratio - to hold high quality liquid reserves assets (HQLR) in amounts that allow them to pass liquidity stress tests.  Such assets are short term TBills and reverse repos, etc.  If you want more info on what are considered HQLR assets, see here.

So, even though "reserves" as such are no longer required it doesn't mean that banks can operate at full-risk-tilt.

 

Canary In The Mine

 Back to sudden debt.. lots and lots of zooming debt, and potentially in lots of trouble.

Open AI has borrowed $96 billion up to now, with major lenders being SoftBank, Oracle and a bunch of private debt funds. But, it turns out that Open AI is being rapidly overtaken by Google - not exactly a surprise, given Google’s massive data supremacy.

Taking a look at SoftBank and Oracle stocks… they are down sharply from their recent bubbly tops: SoftBank is down 30% and Oracle 45% while our blast-from-the-past 5 year CDS (Credit Default Swap) for Oracle has gone from 45 to 125 recently. Likewise, SoftBank CDS jumped from 200 to 300 in the last two months.  Note that Oracle's debt is rated BBB and Softbank's BB+, not exactly stellar ratings.

And there is a lot of AI credit exposure to shadow banking out there, too.  But just because it’s in the shadows doesn’t mean the “real” banking world is immune. In the end, all banking is connected, one way or another.