Why one layer pulled down the Nasdaq-100.
Last week we looked at the Nasdaq and discovered that QQQ isn’t the Nasdaq. This week, we are going one layer deeper – because “technology” isn’t one investment either.
Technology isn’t one sector. It is a supply chain.
Let’s follow a single request through the technology stack.
You type a question into an AI program. The answer appears almost instantly, but underneath, there is a lot going on. Like the duck gliding serenely across the water while paddling frantically below.
One click. Seven layers.
One click sets the entire stack in motion:
- Chips calculated it.
- Servers house the chips.
- Data centres house the servers.
- Networks move the data.
- Cloud platforms run it.
- Software delivered it.
- Cybersecurity protected it.
That explains why buying “technology” can mean buying very different businesses.
We all know that billions have been committed to AI.
But where has the money actually gone?
This week, rather than examine all seven layers, we’re going to compare the two ends of the AI story: the companies that built the machinery and the companies expected to use it.
Where the money has gone
The Building Blocks: Semiconductors
The first AI money went into building the machinery. Semiconductor indexes soared -until good results were no longer enough. They were expected. Expectations had risen even faster than earnings.
As we have seen recently, semiconductor companies could produce excellent results, beat analysts’ forecasts hands down—and still be punished. Once expectations become high enough, “excellent” can be disappointing.
There are numerous semiconductor indexes and ETFs. Here are three of the best known:
They had been on a tremendous bull run – until they came back to earth at the pink dashed line.
A rebound? Possibly. The chart might support it.
In actuality? Who knows?
Where the market hasn’t followed
The users: SaaS
Software-as-a-Service companies were expected to be major AI beneficiaries. But many faced the opposite problem.
The cost of adding AI was immediate. The software companies still had to prove that customers would pay enough for it to justify what had been spent.
The closest investable measure I can find is WCLD, the WisdomTree Cloud Computing Fund. It tracks the BVP Nasdaq Emerging Cloud Index, which currently contains around 66 companies providing cloud-based software and services.
WCLD performed spectacularly during the COVID period, as did many other technology companies such as Zoom. Since then, however, it has been far from stellar.
Here WCLD is compared with SOXX, an ETF that tracks a broad index of semiconductor companies. Both are shown on a percentage scale.
Since ChatGPT was launched at the end of 2022, SaaS has essentially flatlined while semiconductors have taken off.
The builders were paid first. The users still had to prove that the investment would pay.
Why the Nasdaq-100 has been falling
The Nasdaq-100 holds companies from across the stack—but not equally. Several of its largest semiconductor holdings currently exert far more influence than its numerous, much smaller software businesses.
Weights as at 30 June 2026.
Semiconductor Stocks
The four highlighted semiconductor companies – Nvidia, Micron, AMD and Broadcom – accounted for approximately 19% of the Nasdaq-100.
When investors began questioning the likely return on the enormous sums being spent on AI, these heavily weighted semiconductor shares fell – and pulled the entire Nasdaq-100 down with them.
SaaS Stocks
SaaS, meanwhile, has been quietly left behind.
WCLD is broadly spread across dozens of emerging cloud companies. At 30 July, its largest holding, Datadog, represented just 2.82% of the fund. Palo Alto Networks was next at 2.66%.
Even these companies have little influence on the Nasdaq-100. Palo Alto Networks, for example, represented only about 1.2% of the index.
Although Microsoft, Alphabet and Amazon operate enormous cloud businesses, WCLD targets emerging companies whose primary business is cloud software. It does not hold the established technology giants.
What happened during July
Semiconductor shares had enjoyed a tremendous run – and then reversed as investors questioned whether the massive expenditure on AI would produce adequate returns.
Because chip companies now carry so much weight, their falls pulled the Nasdaq-100 down with them.
Then came the sharp rebound.
Microsoft reported 43% growth in Azure. Amazon reported 37% growth in AWS – its fastest cloud growth in more than four years. Investors were suddenly reassured that all that AI expenditure might indeed be producing results.
The Nasdaq-100 was not simply “technology going down.”
It was the market repeatedly reassessing which layer of the AI stack would make the money – and when.
One label. Very different businesses.
The builders were paid first.
The users still have to prove that the investment will pay.
The lesson is to look beneath the “technology” label. Which part of the stack are you actually looking at?
To the markets . .
A mixed week – but look on the bright side: we ended the week higher than last week. Not a lot, I admit. But in this sideways – or consolidating – market I’ll take all the good news I can get. I must say I find consolidation periods extremely tedious – but then when we go into a downward slope I long for sideways!
SPY Charts
Wednesday was a particularly nasty day – a big red candle heading towards the support line (pink dashes). It did not look good. However, Thursday and Friday were more positive and on reasonable volume.
Longer term, SPY seems to be using the upper boundary of the trading channel as a support line. I am not complaining in the slightest.
SPYG Charts
Last Wednesday did not look good on SPYG either: heading down towards the previous high. But Thursday and Friday saw upticks – and closed above the 10-day SMA for the first time in a while. Clutching at straws? Very possibly.
Longer term, SPYG is in the middle of its trading range, and you can clearly see the sideways movement.
QQQ Charts
And the one that has been giving us all a headache! Last week we were thinking that we might be heading for a death cross – well, we have had a last minute reprieve!
Wednesday was awful, but Friday saw us back at the support level—although still on the wrong side of it. If QQQ breaks back through, it will be a good sign. If it falls again, the previous support may become resistance. Let’s hope not.
Longer term, we can see how serious the dip was – or is. If the trading channel is to hold, QQQ can’t drop much further. Let’s hope it does hold.
VIX Chart (Volatility)
The VIX seems unperturbed.
ITMeter
The week ahead . . .
This week’s focus shifts to the US labour market, with employment reports building towards Friday’s crucial jobs figures. Manufacturing and services activity will also be closely watched, while results from Palantir, AMD, Eli Lilly, Uber and Disney could produce further volatility across technology and the broader market.
Monday
- Reports: Manufacturing PMI, ISM Manufacturing, Construction Spending
- Earnings: Palantir, ON Semiconductor, Marriott, Tyson Foods
Tuesday
- Reports: US Trade Balance, JOLTS Job Openings, Factory Orders
- Earnings: Caterpillar, McDonald’s, Pfizer, Merck, Spotify, Shopify, AMD, SpaceX
Wednesday
- Reports: ADP Employment, Services PMI, ISM Services
- Earnings: Eli Lilly, Uber, Disney, SanDisk, CVS Health
Thursday
- Reports: Weekly Jobless Claims, Productivity & Labour Costs, Wholesale Inventories
- Earnings: ConocoPhillips, Fiserv, Airbnb, Cloudflare, Atlassian, DraftKings
Friday
- Reports: Nonfarm Payrolls, Unemployment Rate, Average Hourly Earnings
- Earnings: Allianz, Munich Re
The futures . . .
The headlines are bad:
But the futures don’t seem to have read them:
Heather
Trade the tide, not the waves
Q & A
Sorry everyone, I’ve got a bit behind with the comments. I will transfer anything that I didn’t answer last week to this week and answer them here.
One Response
H- So my blood shot eyes see that the SPY is still in a range… and the other two ( this is where the crystal would help…). I have the chatGPT plus ( an extra 20.00 a month ) – I feel like I’m talking to a 10 year old Rocket Scientist. I have to remind myself it’s machine learning or I’ll pick it up and throw it across the room♂️. R
Ps I just want my personal Ai robot and my self driving car ( or personal flying drone ).