Is Nvidia now the most important share on the planet?

[TheMotorcycleBoy]

Yeah. This has gone bonkers for me.

I topsliced about 15% of my holding a year or so ago when it was +60%.

I'm now staring at a +400% gain, with this now being the largest holding in our folis at 5%.

I guess I'll just hang on for now then.....

[Lootman]

Yeah. This has gone bonkers for me.

I topsliced about 15% of my holding a year or so ago when it was +60%.

I'm now staring at a +400% gain, with this now being the largest holding in our folis at 5%.

I guess I'll just hang on for now then.....

It is hard to sell a share that seems to go up every day. Since just last week my position has gone from being up five-fold to being up six-fold. But it is still only about 2% of my portfolio and a neutral holding is 3%, so I am actually under-weight still!

[TheMotorcycleBoy]

I know what you mean. My NVDA is just over 3x the size of my median holding.

I guess I'm "obese NVDA". But hey.

[SalvorHardin]

I had a similar performance and issues with Soco in the early 2000s. When Soco got to about twenty times what I had invested over the years it was just over 50% of my portfolio (my Soco holding was then worth considerably more than my house). Soco worries were starting to interfere with my sleep.

One morning I woke up, and my subconscious had clearly decided that I should take some money off the table. I sold a quarter of my Soco shares that morning and reinvested mostly in "boring" stuff (consumer staples, distilleries, National Grid and international investment trusts). Problem solved.

The psychology can get a bit weird in situations like this. Fear of missing out on further gains versus Loss Aversion is an interesting battle.

[TheMotorcycleBoy]

Yeah definitely. I guess I'm too busy with other stuff to play around too much with our setup.

If it hits $820 ish again perhaps I'll sell some. But yeah, only to put on boring things instead.

[Lootman]

Yeah definitely. I guess I'm too busy with other stuff to play around too much with our setup.

If it hits $820 ish again perhaps I'll sell some. But yeah, only to put on boring things instead.

In that situation you could sell $820 call options. The premiums are high because of the perceived volatility. So either you get to keep that fat premium (rinse, repeat) or else you sell at your target price. It is like being paid to maintain a limit sell order.

The problem is that options contracts are in 100 share sizes, so you need 100 plus shares, which is a big dollar value. I wish Nvidia would do a stock split.

[TheMotorcycleBoy]

Yeah definitely. I guess I'm too busy with other stuff to play around too much with our setup.

If it hits $820 ish again perhaps I'll sell some. But yeah, only to put on boring things instead.

In that situation you could sell $820 call options. The premiums are high because of the perceived volatility. So either you get to keep that fat premium (rinse, repeat) or else you sell at your target price. It is like being paid to maintain a limit sell order.

The problem is that options contracts are in 100 share sizes, so you need 100 plus shares, which is a big dollar value. I wish Nvidia would do a stock split.

TBF I'm not sure that my iWeb ISA can handle option contracts.

Matt

[Lootman]

In that situation you could sell $820 call options. The premiums are high because of the perceived volatility. So either you get to keep that fat premium (rinse, repeat) or else you sell at your target price. It is like being paid to maintain a limit sell order.

The problem is that options contracts are in 100 share sizes, so you need 100 plus shares, which is a big dollar value. I wish Nvidia would do a stock split.

TBF I'm not sure that my iWeb ISA can handle option contracts.

I know nothing about iWeb but, yes, probably not possible in any ISA. Buying options might be OK but selling options needs a margin account, even if 100% covered.

[TheMotorcycleBoy]

Another matter making me more reserved about sales of any of my US tech shares is murmurings such as:

https://www.reuters.com/markets/us/feds ... 024-03-01/

surely falling discount rates can only push US equities ever higher?

[TheMotorcycleBoy]

Actually looking at my sharepad figures:

reported PE = 66
forecast PE = 32

Period         Q1    Q2      Q3      Q4
PAT         2043   8232   17475   29760

I'm currently thinking that the price comes with some justification, should this growth continue.

[Adamski]

It's just dot com all over again.

In March 2000 it was Cisco making hardware for the new thing, the Internet.

The parallels are striking. The share price assumed Co's would keep buying chips at the same exponential growth rate. Of course this slowed up, and the stock price crashed. In hindsight it proved to be a massive bubble.

Cisco is still one of thecworlds largest tech co's but never recovered its price back to 1999-2000 levels.

Exact same thing today with AI.

Exact same thing with Tesla. Massive bubble, reality will eventually catch up. Once sales slow these will crash.

[SalvorHardin]

Another matter making me more reserved about sales of any of my US tech shares is murmurings such as:

https://www.reuters.com/markets/us/feds ... 024-03-01/

surely falling discount rates can only push US equities ever higher?

The same thing is happening with REITs, where there have been some substantial share price rises in recent weeks on the assumption that interest rate cuts are happening sooner rather than later.

That said AI doesn't seem to be the cure all that some claim it to be. There have been quite a few instances where lawyers used AI to generate arguments in court and it turned out that the AI has used "historical" court judgements which turn out to be completely made up. Google's Gemini AI, in a classic case of "Garbage In, Garbage Out", has increasingly been pumping out nonsense and outright lies and publishing them as fact.

The most high profile (and ridiculed) instant was making every white historical figure non-white. It might be a bit of surprise to find out that according to Gemini Hitler's Waffen SS were entirely black, as were America's Founding Fathers. Gemini has turned into what some lawyers are calling "Google's AI powered libel machine", generating libels at will and Google should be on the hook as a defendant because it exercises editorial control over what Gemini publishes.

The system is in a positive feedback loop when it comes to generating nonsense, and is highly contagious. Other AI programs are drawing upon existing articles which contain numerous lies and generating further libellous articles which in turn are picked up by AI programs scraping the web.

Freelance journalist Matt Taibi (quite well known for reporting for Rolling Stone), sasked Gemini AI "What are some controversies involving Matt Taibbi?” and found a lot of commentary about articles which he didn't write, but were made up by the AI.

With each successive answer, Gemini didn’t “learn,” but instead began mixing up the fictional factoids from previous results and upping the ante, adding accusations of racism or bigotry. “The Great California Water Heist” turned into “The Great California Water Purge: How Nestle Bottled Its Way to a Billion-Dollar Empire—and Lied About It.” The “article” apparently featured this passage: "Look, if Nestle wants to avoid future public-relations problems, it should probably start by hiring executives whose noses aren’t shaped like giant penises...."

"...God knows what Gemini did in my case, but if caricatures of me riffing on Jews with penis-noses are what come out when Google’s “creative tool” runs my name through its Rube Goldberg machine, it’s hard not to wonder what lunacies go on in products like Google search for people generally. The potential for abuse is mind-boggling and almost makes you wonder about the reasons Google released this flawed product."

https://www.racket.news/p/i-wrote-what-googles-ai-powered-libel

Good news for database providers which have curated databases and don't take everything published on the internet to be a credible source

[mrodent]

Haha, one of my favourite rants. The acronym is wrong: it shouldn't be "AI" but "SP", statistical processing. Sometimes the results can be spectacular, such as Google Translate, which is getting better and better particularly for the major languages. But it still fails between English and French, say, in specialist fields, like law or accounting. That matters!

I also fully expect "SP" to deliver better and better autonomous driving. But one thing that will not be involved in that is intelligence, of any kind.

There are various people out there who are saying that the Magnificent 7 rise is not due to this "SP" and the hype surrounding it, and they adduce various statistics about the fundamentals saying why it is not like the dotcom bubble. I'm unable to judge these things, but surely if not a bubble they must at least start to level out imminently.

Personally I'm "tilting" towards small-cap and mid-cap for a laugh. Not timing the market of course just, ahem, diversifying risk.

[simoan]

I had a similar performance and issues with Soco in the early 2000s. When Soco got to about twenty times what I had invested over the years it was just over 50% of my portfolio (my Soco holding was then worth considerably more than my house). Soco worries were starting to interfere with my sleep.

One morning I woke up, and my subconscious had clearly decided that I should take some money off the table. I sold a quarter of my Soco shares that morning and reinvested mostly in "boring" stuff (consumer staples, distilleries, National Grid and international investment trusts). Problem solved.

The psychology can get a bit weird in situations like this. Fear of missing out on further gains versus Loss Aversion is an interesting battle.

I find these days I sleep very well as I no longer run such a concentrated portfolio. I am now much more diversified and risk management is everything now I am retired.

With regard to the risk of being very overweight in US semiconductor stocks I’d ask yourself what would happen if you woke up one morning and China had invaded Taiwan? You need some kind of insurance because without Taiwan, fabless semiconductor companies like Nvidia, AMD and Broadcom are pretty much dead in the water. Yes, everything else would tank too, but such direct exposure to Taiwan would result in huge losses.

[TheMotorcycleBoy]

It's just dot com all over again.

In March 2000 it was Cisco making hardware for the new thing, the Internet.

The parallels are striking. The share price assumed Co's would keep buying chips at the same exponential growth rate. Of course this slowed up, and the stock price crashed. In hindsight it proved to be a massive bubble.

Cisco is still one of thecworlds largest tech co's but never recovered its price back to 1999-2000 levels.

Exact same thing today with AI.

Exact same thing with Tesla. Massive bubble, reality will eventually catch up. Once sales slow these will crash.

However a lot (most?) of the dot.com firms weren't profitable. Some weren't even making sales.

But NVDA is churning out masses of profit, and its grown a lot of late. See my earlier post.

viewtopic.php?p=650360#p650360

[TheMotorcycleBoy]

That said AI doesn't seem to be the cure all that some claim it to be. There have been quite a few instances where lawyers used AI to generate arguments in court and it turned out that the AI has used "historical" court judgements which turn out to be completely made up. Google's Gemini AI, in a classic case of "Garbage In, Garbage Out", has increasingly been pumping out nonsense and outright lies and publishing them as fact.

I totally agree. I'm in computer s/w myself, and of the view that none of this AI, is "intelligence" as such, it's such fast processing combined with pattern matching.

[simoan]

That said AI doesn't seem to be the cure all that some claim it to be. There have been quite a few instances where lawyers used AI to generate arguments in court and it turned out that the AI has used "historical" court judgements which turn out to be completely made up. Google's Gemini AI, in a classic case of "Garbage In, Garbage Out", has increasingly been pumping out nonsense and outright lies and publishing them as fact.

I totally agree. I'm in computer s/w myself, and of the view that none of this AI, is "intelligence" as such, it's such fast processing combined with pattern matching.

As someone that implemented a 16 node neural network in hardware 30 years ago, this isn’t really true. It’s much more than pattern matching. Once properly trained it is capable of making better decisions than humans can in a number of applications. The problem is in training in a way which is free from bias to avoid the same mistakes humans make. As such, any AI implementation that is trained using data from the internet will be worse than useless. Personally I don’t understand all the hype.

[TheMotorcycleBoy]

I totally agree. I'm in computer s/w myself, and of the view that none of this AI, is "intelligence" as such, it's such fast processing combined with pattern matching.

As someone that implemented a 16 node neural network in hardware 30 years ago, this isn’t really true. It’s much more than pattern matching. Once properly trained it is capable of making better decisions than humans can in a number of applications. The problem is in training in a way which is free from bias to avoid the same mistakes humans make. As such, any AI implementation that is trained using data from the internet will be worse than useless. Personally I don’t understand all the hype.

Surely the "better decisions" are based on better matching of inputs against logic trees that have built based on, as you say, training?

There's no gut feel implanted there, or quite literally the ability to "think out of the box".

[simoan]

As someone that implemented a 16 node neural network in hardware 30 years ago, this isn’t really true. It’s much more than pattern matching. Once properly trained it is capable of making better decisions than humans can in a number of applications. The problem is in training in a way which is free from bias to avoid the same mistakes humans make. As such, any AI implementation that is trained using data from the internet will be worse than useless. Personally I don’t understand all the hype.

Surely the "better decisions" are based on better matching of inputs against logic trees that have built based on, as you say, training?

There's no gut feel implanted there, or quite literally the ability to "think out of the box".

That’s the whole point though. “Gut feel” is a horribly human concept that often leads to poor decisions being made and when improperly trained, AI will do the same. AI is not based on logic, it’s much more fuzzy than that i.e. a multi-layered interconnected network of neurons each with weighted outputs based on a set of weighted inputs. The problem is, the AI can’t say “I don’t know what to do” like a human, it will make a decision no matter what. This is why the current generative AI tools trained on nonsense off the internet just make up facts.

[ignotus20]

However a lot (most?) of the dot.com firms weren't profitable. Some weren't even making sales.

But NVDA is churning out masses of profit, and its grown a lot of late. See my earlier post.

viewtopic.php?p=650360#p650360

The dotcom comparison is with Cisco, who were profitable and who provided the infrastructure to support the existence of dotcom firms. This article is interesting: https://www.kingswell.io/p/the-rise-and-fall-of-cisco

Simoan's point about China and Taiwan is a good one, also made here: https://www.ft.com/content/bec85749-935 ... 23541c7bce

The implications if Donald Trump becomes president again add some more chaos into the mix: https://asia.nikkei.com/Politics/U.S.-e ... -on-Taiwan

This seems like a stock to trade, as opposed to invest into currently - and with a tight stop.