AI: Bubble or Hype Cycle?
Tariffs and market volatility are starting to become the new normal; market fluctuations have become yesterday’s news. Instead, we have the return to the AI ‘bubble’ and dire warnings about following the ‘tech wreck’ of 25 years ago.
But is it a bubble?
Market bubbles occur when there is a rapid rise in the price of an asset driven by speculative demand followed by a sharp collapse in prices. There have been bubbles since Roman times (Republic Land Bubble 150-33 BCE), but Tulipmania was probably the most bizarre.
Tulipmania
In the 17th century global trade started in earnest, and flower bulbs from Asia like tulip, crocus and hyacinth became luxury items. Wealthy Dutch started to cultivate tulips and speculation in bulbs began.
In late 1636 frenzied trading began, often involving futures contracts. People were buying bulbs, not to plant but to resell at a profit.
Market Collapses
By January 1637 prices reached their peak and rare bulbs traded for the price of a house.
In early February an auction failed to attract buyers, and the market collapsed abruptly.
Within days prices plummeted, and many bulbs lost over 90% of their value.
Is that what's happening now?
Here’s a chart * published last week in Investing.com likening the current situation to Tulipmania. The author’s assessment (not mine) of where we are now in the AI bubble is the blue arrow.
In other words, we are now in the mother of all bull traps, and the market is going to plummet further. But is there another way of looking at things?
The Hype Cycle
Those of you who have been in IT for a while will be very familiar with the Gartner Hype Cycle which postulates that all new technology goes through the same stages (with very flamboyant names!)
If we apply the hype cycle to the development of the Internet then we would have:
- Innovation Trigger (1970s / 80s)
- Peak of Inflated Expectations (1994/2000 – Dotcom Boom)
- Trough of Disillusionment (2000/2002 – Dotcom Bust)
- Slope of Enlightenment (2003/2010 – survivors mature, social media)
- Plateau of Productivity (2010 – present – now ubiquitous)
Early AI
We’ve been here before, but thhe first wave of AI died a quiet, unnoticed death. Then it was called ‘expert systems’ – if you have been around IT as long as I have you will remember them. They reached the trough of disillusionment in the 1990s but never reached the slope of enlightenment and survived only in niches.
Modern / Generative AI
Starting in 2006, things started to move. Remember Big Data? The Internet of Things? That was stage 1, the innovation trigger. By 2023 we had reached the Peak of Inflated Expectations, with a huge investment boom, the ‘Magnificent Seven’, public hype, media saturation and unrealistic business expectations.
Trough of Disillusionment 2024 –25
We know that only too well. Widespread reporting on AI-generated errors (OK, I’m guilty!), bias, legal and regulatory challenges, and disillusionment by enterprises as generative AI fails to deliver promised productivity gains.
Plateau of Productivity (2028-32?)
This will happen when AI becomes the invisible infrastructure behind most digital systems, much like the Internet is now. When people stop talking about AI, when it is assumed that it is embedded in tools, like Microsoft 365 and Salesforce. When generative AI powers content creation, analytics and automation.
The Slow Death of Search
Funnily enough, after I discovered ChatGPT in late 2022 I thought it was a game-changer. I was so impressed that I actually registered some URLs. They were mine for a few weeks and I was thinking about how I should develop them. Then, abruptly, they were cancelled. Google must have thought I was a real threat!
Currently, google still handles 90% of global search traffic and is likely to dominate for at least a couple more years. But I think change is happening.
For example, I no longer use Google unless I am looking for a particular website. For anything else it is AI, especially when its a tech query – I can’t remember the last time I read an online manual. WordPress, Elementor, Windows, Excel, Clipchamp, even Schwab – my first stop is AI.
For example, for a tax return yesterday I needed my HIN number (an Australian thing) and I was poking unsuccessfully around NABTrade website trying to find it – then just asked AI where it was – et voila! The answer.
I’m sure as other people realise the time that can be saved, adoption will be swift, especially as Google is becoming more unusable by the day. And Google has realized it too, planning to invest $75 billion this year in expanding its data center capacity.
However, regulatory threats remain: The Dept Justice is bringing antitrust actions, and Google may be forced to get rid of key assets like Chrome.
So – Bubble or Hype Cycle?
It is only my opinion, but I would suggest that we are not in a bubble, but in the trough of a hype cycle. I do think that AI will rise again, and with it the tech stocks that are most involved. But happy to hear arguments against!
Here’s a chart from The Economist:
To the markets
The market has been much more positive this week, although not definitively bullish. For now the specter of the bear has receded, and we are all hoping that another bull market is starting.
Well, I definitely am; while bear markets are exciting, they are also fast moving and need closer attention – which is something I don’t want to do when I am on holiday.
Yes, I’m off on my summer wanderings, and won’t be home until August. Exciting – but while I will be keeping an eye on the market, I may have some problems, especially in the Arctic Circle. Goodness knows what the comms are like up there!
SPY Charts
Since the huge green candle on the 9th April the trading has all been within its range – until last Friday, when it closed just over 550. You can see that an ascending triangle is being formed, and charting wisdom is that a break to the upside is a bullish signal. I don’t think that Friday was a very convincing break to the upside. It has to be a bit bigger and on higher volume before it qualifies as a real signal, but it is encouraging less.
The big news is, of course, that it is bumping up against the correction level.Another up day and it will be over it. You can also see that it didn’t spend any time in bear territory, but simply used it as a support level. Amazing how accurate charts can be!
We can also see that the 10 day SMA has turned upwards again – so, of course, we are all cheering it on and hoping for a golden cross. It looks as though it may happen in the 570 range (remember that the 200 SMA responds much slower than the 10 SMA) so possibly we will be buying in at around the same level that we got out. We’ll see.
The longer term chart was a bit of a surprise. I haven’t updated the charts on this laptop since my travels ended in August last year, so I was curious to see what had happened with the lines I had drawn in. You may recall in previous blogs I was somewhat nonplussed at the level SPY was bouncing as I couldn’t make sense of it; well, looking at the longer term chart it all becomes clear!
SPYG Charts
SPYG is following the same path as SPY, although it has closed more significantly above it. If it continues up it is likely that the golden cross will be in the 82 – 83 level, which is also the correction threshold. Let’s hope.
On the longer term chart, we can see the support at the previous high in 2021 held and it bounced higher.
QQQ Charts
QQQ has pulled away from bear territory and is heading towards getting out of correction territory also. Like SPY, it has started to trade above the level of the big green candle on 9th April.
On the weekly chart we can see that it has bounced off support.
VIX Chart (Volatility)
The VIX is settling down, but we’d still like it a bit lower.
ITMeter
Chat & Charts
I don’t keep tabs on what is happening in chat as I don’t want people to think that I am snooping on them – you know that if you want to ask me about something that the comments or the contact form is the best way. I did, however, notice in the database that people were having trouble locating the charts on the website.
The word ‘charts’ was use both as a heading and a link, which confused people, but I have now moved charts under a new menu item of Tools, so I hope that is easier.
The week ahead . .
Last week Alphabet (GOOGL) reported better-than-expected results, which was good news for the tech stocks. There are some big tech earnings due this week – AAPL, MSFT, AMZN and META. There is also big oil: XOM (ExxonMobil) and CVX (Chevron) are reporting, as is Visa (V), Caterpillar (CAT) and Coco-cola (KO).
It may be a volatile week!
Futures
The futures are down a bit, but it is still 10 hours to market open.
Today’s blog brought to you from . . .
An annoying little laptop. I travel as light as possible, which unfortunately means a nasty little keyboard and a nasty little screen.
It has taken me twice as long as usual to write and put up this blog, so they may be a bit shorter for the next few months.
I’ll make sure that I cover the charts and any signals but maybe keep the commentary to the minimum.
We are still out of the market, but let’s hope.
Fingers crossed for a golden cross!
Heather
Trade the tide, not the waves
8 Responses
Hi Heather,
Thank you for your work — I’ve really been enjoying your blog.
I noticed that some of the charts use the SMA 10 and SMA 200, and I was wondering: for the golden cross you mention, are you specifically referring to the SMA 10 crossing above the SMA 200?
Is there a particular reason you chose this combination? I’m curious whether it is based on historical data, and if you explored other combinations, such as SMA 20 and SMA 200, or perhaps EMA 20 and EMA 50. I’m very interested in learning which pairing has historically produced the best results.
Thank you again for sharing your insights!
Best regards,
Tyler
Hi Tyler,
You need to buy heather’s book, In The Money, Bull Market. It explains everything.
Thank you Jim!
Hey Tyler
I explored every combination of indicators I could think of, then every combination of parameters I could think of. Many many months of work.
Surprisingly, the 10/200 SMA came out tops – I say surprisingly as I would have been happier if it had been somthing more arcane – how much more impressive would it have been using the 7/183 cross?
But unfortuanately the 10/200 beats the other hands down ( although only just beating the 10/199)
If you check the backtesting link it gives you the method of creating your own backtesting systme so that you can check for your self – whcich is a good idea, you would feel more comfortable using it.
Hope this helps
h
No, but thinking of a trip there maybe in two years. Have fun and don’t feed the polar bears 🙂
well, I certainly won’t feed them with ME!
Will post photos, so looking forward to it!!
h
Will you be going to Svalbard Heather? Comms should be okay.
Early May is usually an interesting time for the markets; I remember some devastating plunges but we shall see.
Hey Roger – yes – Svalbard is on the agenda! So very excited, have never been that far north before.
You don’t live there, do you ?
h