Progress at a price

Artificial intelligence (AI) represents one of the finest advancements in modern technology and engineering. It promises us efficiencies and savings across almost every industry, but this progress comes with a price. The trouble with a siloed world, where production is separated from product, is that we don’t see how things are made. Be it the clothes we buy - made who knows where and by whom - or the AI we have unthinkingly incorporated into our day to day, everything comes with a cost; we just may not be able to see it.

But these costly products are either essential to our lives or add richness in other ways. AI may come with a price, but it brings creativity and enormous efficiency, as illustrated here by ChatGPT’s rhyming expertise. We asked the programme to write a poetic introduction to a quarterly investment update and were amused with the results, which it produced in two seconds:

Possibly not CapGen’s next formal investment philosophy, and admittedly not solving any global issues, but impressive nonetheless in its immediate innovation. AI presents the world with challenges, but it also undoubtably adds new layers of imagination and intellect. Progress, at a price.

Artificial intelligence (AI) is predicted to make productivity savings across almost every industry in the world. But savings often represent cost somewhere else. AI may be able to create efficiencies, but this artificial thought requires more electricity than the world can currently generate. McKinsey forecasts a 10-12% annual increase in global data centre power demand between 2020 and 2030. Where will this energy come from?

Fossil fuels offer the quickest solution, which means the cost to AI’s productivity savings will be to global clean energy efforts. Currently, renewables simply cannot offer the immediate relief that fossil fuels can.

Wind and solar do provide a partial solution but their weather-dependent nature means alternatives are needed during adverse conditions. Storing sustainably generated electricity also presents challenges, as excess energy cannot be stored without massive battery arrays, which come with their own sustainability costs.

So, eyes turn to fossil fuels, and natural gas looks attractive as the construction window for a combined gas turbine is roughly 30 months. But nuclear is expected to win the race, due to its carbon-free credentials and dispatch profile. AI certainly doesn’t push things in a more renewable direction.

There are also some practical and social issues with the AI data centres, through which all this energy is directed. While data centre development is valuable to global GDP, it puts pressure on local electricity and water use. New data centres will also face equipment and construction bottlenecks in direct competition with utilities and industries. On top of that, they provide little ongoing local job creation, while tax revenues often end up in offshore jurisdictions.

AI does undoubtably have the potential to drive vastly improved energy efficiency, and ultimately lower the world’s electricity needs. But to get to that point, there are plenty of hurdles to overcome.

The S&P 500 has had a strong six months, returning almost 15% year to date (as at 30 June 2024), and frequently closing at record highs. But take a closer look at this stellar performance, and some investors get a little nervous. Most of the performance can be attributed to just five big names.

Of the S&P 500’s overall 15% return, c. 60% has come from five stocks (or just 1% of the index): Nvidia, Amazon, Meta, Microsoft, and Apple. Drill down even further, and actually, semiconductor manufacturer, Nvidia, alone drove almost a third of that overall 15%.

As the S&P 500 is weighted by market capitalisation, the bigger a company’s valuation, the bigger its representation in the index. Nvidia’s share price has risen by nearly 150% since the start of the year, sending its valuation past $3 trillion. With a market-cap of that size, Nvidia now dwarfs most other businesses in the S&P, only vying with Microsoft and Apple for top spot. The combined value of these three stocks now represents approximately 20% of the index, making it very concentrated at the top.

The S&P’s headline performance may look great, but this heavy reliance on just a few tech behemoths certainly represents a big risk. If Nvidia et al’s performance continues to soar, then it’s a wave that every investor will want to ride. But if their performance starts to falter, then the S&P will be vulnerable to a hard hit.

We can’t help but remember the dotcom bubble’s Cisco Systems. The networking hardware firm was briefly the most valuable company in the world, before losing c. 80% of its market capitalisation in two years, as the dotcom bubble burst. There are plenty of similarities between Cisco and Nvidia which make us wary of this extreme market concentration. But then again, perhaps Nvidia’s business model really is that good. Only time will tell.

Copper is a versatile metal, used in building construction, electronic product manufacturing and in various household appliances. But it has another use that flies slightly under the radar; historically, it’s been a strong leading indicator of global demand. It could be said that copper has a PhD in Economics. So, what is Dr Copper telling us with its 13% price rise year to date (as at 30 June 2024), and where has it come from?

Like most market movements, it's down to supply and demand. We believe we are late in this economic cycle, which is typically a good time for commodities as their prices rise with inflation and they offer diversification from increasingly volatile late cycle stocks. Commodities have felt the benefits of today’s late-cycle driven demand increase, which is one of the reasons behind the metal’s positive performance this year.

Longer-term increased demand has been driven by electric vehicle (EV) production, as copper plays a major role in electric motors and batteries. The move towards a greener economy has pushed up the structural demand for copper more than ever before.

On the supply side, there’s been an underinvestment in copper mines. It takes a long time to respond to these shortages, which is the reason we see long boom and bust cycles in commodities.

All in all, we think the cause of copper’s price increases can be attributed to short-term excess demand and longer-term increased demand, combined with a supply shortage. Does this tell us anything meaningful? Many famous speculators are watching copper closely and buying it for those short- and long-term economic reasons. We recognise the sentiment around copper and think of it as a good signpost, but we’re no longer convinced of its intrinsic value as a leading indicator, as its renaissance is so linked to EVs.

If you’re looking for a commodity-based market timing indicator of the modern world, perhaps it’s better to look at semiconductors themselves. Their prices and demand are just as useful for the modern investor as copper was in the past. It’s always important to broaden your toolkit of indicators.

A couple of years ago, we set ourselves three questions. Do we think a recession is more likely than not? Do we think we are in an inflationary regime or not? And if we think there is a recession, how severe or otherwise do we think it might be? At that stage, we thought the risks were balanced. Fast forward to today, and we continue to believe the current environment is inflationary and the possibility of recession remains elevated. But what does that mean for the future? We’ve finessed our views into three possible paths ahead.

The first scenario is recessionary. It says there’s a possibility growth is already slowing, and the slowdown will continue until we reach recession.

There are several data points which support scenario one. Short-term estimates of US GDP have seen a considerable drop-off in the past couple of months. US manufacturing PMIs have been below 50 for 18 months, a level which typically indicates contraction. New orders are often a leader of manufacturing, and although they have picked up a bit, the numbers are still fairly weak. This suggests recent strength could point to inventory restocking, as opposed to genuine recovery.

US unemployment data is mixed. There are signs of weakness, including high initial jobless claims, and some longer-term unemployment measures are starting to pick up. Job openings have slowed, which should lead to a drop in wage pressure, taking the heat off inflation. But falling wages can lead to a slowdown in activity.

US consumer sentiment recently has been poor, yet the economy has generally been surprising to the upside. Most people in the US think they’re in recession, even though they’re not, and that’s partly because of the cost of living. Prices have shot up, but wages haven’t. Consumers are loading up their credit cards and delinquencies (the state of being in arrears) are rising quickly among those most dependent on credit.

There’s plenty of evidence to support scenario one’s continued slowdown, and data have deteriorated over the past couple of months. But the bright spots might continue to shine, so we’re not placing a binary bet on this outcome yet.

Scenario two is also recessionary, but it makes different assumptions about the current state of the economy and route to slowdown. The second scenario says the economy has slowed a little but will remain strong, however, persistent inflation and higher interest rates will eventually lead us to a slowdown. Falling inflation isn’t typically associated with a strong economy, so if the economy is as robust as it could appear, inflation is likely to stay where it is (higher than central bank targets), or even move up.

So, what evidence is there to support today’s economic strength? 2023 saw strong government spending and consumption; certainly, stronger than we were expecting. We’ve seen some of this strength continue into the first quarter of this year, before falling off a little over the past couple of months. Markets are going up and consumers are earning higher rates on their cash. Household balance sheets are in a stronger position, and that’s what led to the consumption binge which has supported the US economy over the past couple of years.

But inflation is elevated, particularly in services. The prices paid within services is a particularly important measure, and that’s been increasing. Service sector inflation puts inflationary pressure onto the wider economy, which means the Fed would then struggle to cut interest rates.

It isn’t just services where inflation is elevated. Pure goods inflation has also been trending upwards over the past three or four months. That means there’s now a dichotomy between what the Fed wants to do and what the market wants the Fed to do, which is why market expectations for interest rate cuts have been completely repriced this year. If the economy continues as it’s moving today, then there won’t be any leeway for the Fed to cut rates. Sticky inflation and higher interest rates will eventually slow the economy right down.

The third scenario paints a more optimistic picture but has the least supportive evidence. Scenario three says we escape a slowdown completely, due to the efficiency and productivity gains realised by AI. Prices will come down, inflationary pressures will dissipate, interest rates will normalise, and economic growth accelerates.

But at the moment, no one quite knows if AI is going to work. Businesses are implementing it in their systems and processes, but we need a couple of years to see whether it really does drive genuine improvements. AI could still save the day, but it could also be an overhyped and expensive disappointment. The jury is still out on scenario three.

We think scenarios one or two are the most likely, because they generally fit with where we think we are in the current economic cycle. We think we are late cycle, and this cycle began with the recovery following 2008’s global financial crisis (GFC). Covid was a brief ‘stop-start’ within this cycle, and not the complete beginning of a new one.

Scenario three suggests we’re at the beginning of a new economic cycle, and that Covid marked the end of the post-GFC era. Scenario three, we believe, has the least evidence to support it. It’s tempting to imagine AI swooping to our rescue, but economics is rarely so exciting. A delicate dance between interest rates, inflation, and growth may not be so headline grabbing, but it’s the tortoise that wins the race.

Capital Generation Partners LLP (“CapGen”) is authorised and regulated by the Financial Conduct Authority and is registered as an Investment Adviser by the US Securities and Exchange Commission. All information and opinions expressed in this article are subject to change without notice and CapGen and its affiliates do not warrant or guarantee its accuracy, reliability or completeness. Nor does this article constitute investment advice, an offer, contract or other solicitation to purchase any assets or investment solutions, or a recommendation to buy or sell any particular asset, security, strategy or investment product. The information contained in this article does not constitute research or recommendations from CapGen and please note that the value of any investments referred to in this article and their income may go down as well as up. Independent advice should be sought where appropriate and no liability is accepted, or responsibility assumed, in respect of persons who are not clients of CapGen, unless expressly agreed in writing. Advice is given and services are supplied by CapGen on the basis of our terms and conditions of business, which are available upon request.