Join The Multi Asset Lab for less than £2 pcm per asset class and receive:

  • Monthly updates on 15+ separate asset classes from best-in-class commentators
  • Free Subscription to the Property Chronicle Magazine - Digital Edition (normal price £50pa)
  • 30 days free trial!

August was the first down month for the S&P 500 Index since February. Given that the market has managed a gain every month since and even including March, the month that Silicon Valley Bank collapsed at warp speed, it was perhaps due a well-earned summer break to rest up before a further push higher as the days shorten. We shall see. However, a loss of approximately 1.5% tells us little of the detail.

Volatility appears to be awakening from months of slumber, although it may not yet be widely apparent. Even the index of volatility itself was volatile – rising and then giving back over 20% during the month. Without any drama from the US economic data releases, no bombshell dropped from the meeting in Jackson Hole, (unlike last year) it might also be puzzling that the yield on the 10-year US Government Treasury Bonds also fluctuated 10% during August. Clearly the tug of war between the inflation and deflation camps is growing stronger, with China a new wildcard after poor growth figures produced the biggest macro surprise of the month. How their domestic property crisis unfolds may have the greatest impact on global growth and therefore inflation.

In terms of individual stock performance for the month of August, a clear shift was seen towards more defensive, less economically driven companies; 4 of the top 10 best performing S&P stocks were related to the Pharma industry, while most of the worst 10 were from Retail and the consumer centric part of the solar industry. The Mega Cap Tech names that have become synonymous with the US market advance took a month off, with profit taking accompanying almost all their earnings releases. A negative weighting rebalancing for both Apple and Microsoft in the Nasdaq helped them to be the weakest of the group.

Is this weakness in consumer stocks the first chink in the armour of the market? Sceptics have been wondering when and some even yearning for the US economy (which is the US Consumer after all) to bow down before the gloomy predictions of recession. The fastest interest rate tightening cycle in history has had little impact thus far because the American consumer started the year in the best shape of its life! Virtual full employment and lowest liabilities for homeowners, who had locked into mortgages at the lowest rates in history (almost all US mortgages are fixed rate, a legacy of their housing crisis in 2008). While there are now signs of weaker spending as shown by these stocks, as I am no economist, I will not join the group who usually predict 14 of the next 3 recessions.

For us at Shard Capital, it is about picking companies and fund managers who can outperform both for the long and shorter term. So, for this first commentary, I thought I would focus on the world’s most talked about company, at least in investment circles, and now one of the largest as well after it has TRIPLED in six months to be worth well over $1tr (yes, trillion). I speak, of course, of Nvidia.
Every fund manager benchmarked against the S&P must surely own some of it by now, or at least consider doing so and if not, they need explain to their boards, colleagues and clients why they do not! Whether by luck or intent, Nvidia’s GPUs (Graphics Processing Units) originally designed for use in Gaming, turned out to be perfect for machine learning as well. As the leader in that field, Nvidia is the early pace setter in the latest “new, new thing”; Generative Artificial Intelligence, brought to prominence only last November with the release of ChatGPT. Nvidia is to AI what Cisco Systems was to the Internet in the 1990s; the “picks and shovels” in this latest goldrush.

Source: Nvidia via WSJ, 2023

The explosion in revenue expectations from one quarter to the next announced by Nvidia in May and then fulfilled in August was astounding and a feat last seen almost 20 years ago, when Google reported its first earnings as a public company. This is because for now they are the only game in town for the semiconductors that Cloud Service providers must buy, in order to run the AI machines that developers of AI applications are rushing to use to produce a product that end customers will clamour to buy. Currently money is no object, supplied by Venture Capital or the Tech Titans, all desperate to find the next winner.

The question now is whether this is sustainable, or a demand bubble, however profitable it may be. For those that can’t remember or weren’t watching stock markets, the dominant theme in technology spending back at the turn of the millennium was all about the year 2000 itself; or “Y2K” for short – a phenomenon caused by, what turned out to be unfounded, fear that all the computers around the world would collapse as the date clock inside turned over from 1999 to 2000 (as only 2 digits had been assigned in the original programming the machine would assume we were back in 1900). Huge amounts of money were spent by everyone from the US Government down to update all critical systems in advance. There was no problem as it turned out, although some naturally say that was because of all that investment. The point is that it pulled forward a huge amount of IT spend on infrastructure into one glorious year for Cisco.

The same could be happening right now for Nvidia for two reasons; the first is China, whose companies are rushing to make purchases of advanced semiconductors before they are banned by the USA. How much of their spending was in Nvidia’s latest number and what will be in the next I don’t know, but it has been reported that $5bn of demand has been received, about half of the figure in the chart above. The second, greater danger to future chip sales is talk of a ‘pause’, while we work out a framework of exactly who, how or what is going to be allowed to produce services based around AI that regulators and ethics will permit; already there is talk of limiting the actors allowed to operate machines or even buy GPUs for these new forms of intelligence. Oh yes, and how it is going to be paid for? – this stuff is very, very expensive mainly due to the power consumed. For now, no one is worrying about that: the assumption is “build it and they will come”, all very similar to the Dotcom era.

Then there is competition. If all you can afford is a team of horses, you can still plough and sow a field rather than use the latest self-driving tractor from John Deere. In the same vein, there will be other, cheaper versions of what Nvidia is doing today. It is just ludicrous to think otherwise – life doesn’t work like that, especially in technology with plenty of other hungry experienced players. 

To be fair, not owning Nvidia for its AI exposure over time might be the same as not owning Google in hindsight, if you knew what “Search” would do for the world. But timing and price are everything. Unless you are aboard as a very long-term investor, the advice we gave was to take profit after these results last week (the stock opened over $500) as we feel that to extrapolate from current demand might well be wrong. If that proves to be right, it might cost you 25-50% from these levels. If that does happen then for those uninvested, that could be a chance to buy. In the meantime, there are plenty of other much cheaper and overlooked companies that are benefiting from this theme; Dell Computers for one – you need something to put that GPU in after all, it’s called a Server. Have some other ‘fish’ to go with your ‘chips’.

Join The Multi Asset Lab

Only £1 per month per asset class | 30 Days Free Trial!