November Investment Views
View from the Road
September and October tend to be a busy time of year. Market participants return from the summer to what tends to be a seasonally volatile period. It’s a time with a lot of analyst meetings, investor presentations, and conferences.
Following a strong run for equity and fixed income markets, investors have been considering whether recent trends will continue into the end of the year, or in a world where the news flow moves so quickly, new surprises await.
Technology and politics have been key drivers of news flow and financial markets this year and a fractured geopolitical backdrop adds a layer of complexity. These forces are interconnected, with trade seen as a matter of national security. The US reliance on China for critical materials, and China’s interest in advanced technology from the US having been a flashpoint. We are living through a fascinating period in history for the global economic system.
As clients will know, our investment process centres around a range of independent research providers. These specialists focus on specific global regions, asset classes, or individual companies.
While we favour independent research providers, we also follow research from the large Wall Street banks, who have strong research capabilities and access to high quality data. We spend a lot of time blending these views together into a coherent investment strategy.
In recent weeks, we have attended many meetings, including travelling outside of our offshore jurisdictions to get a better sense of the investment climate. This included the BCA Research conference in New York, a flagship event from one of our key research providers.
We thought that it would be helpful to write up some “trip notes”. We do not attempt to answer every question, but frame the investment backdrop and cover key topics which will help define the coming months and years ahead. We hope that you find it interesting.
This is Still an Unusual Cycle
- The pandemic distorted the economy cycle
- These distortions have faded but still evident
- Lower population growth slows job growth
A topic that we have written about in recent years is how the pandemic cycle was truly unique. Investors often think about business cycles, but the pandemic caused a boom in some industries like e-commerce and a major downturn in others, such as travel and leisure. It is striking that five years on, we are still seeing distortions compared to a “normal” business cycle.
Some previously reliable market and economic signals have stopped working. An inverted yield curve (short-dated rates higher than long rates) signalled a recession that never came, as did the Conference Board’s index of Leading Economic Indicators. The “Sahm Rule” showed a reliable historical dynamic that once the unemployment rate starts to rise, it usually continues.
The threshold (three-month average unemployment rate rises by at least 0.5% above its 12-month low) was reached last summer, but the unemployment rate has since stabilised. The labour market is in a historically unusual situation where both hiring and layoffs are both low.
Why the triggering of the Sahm Rule did not lead to a recession remains a key topic. Significant government spending entered into the economy during the pandemic, which saw a huge demand for workers. As economic growth then slowed in 2022 and 2023, demand for workers cooled. The unemployment rate reached a low of 3.4% in April 2023 and has since risen to 4.3%.
Historically a weakening labour market reaches a tipping point where job losses lead to more job losses and a recession. This time, the higher unemployment rate was not driven by job losses, but the labour force growing while hiring was muted. Meanwhile, households had accumulated savings during the pandemic, which helped maintain spending.
Elevated immigration into the US helped lower inflation from the 2022 highs, as it helped keep wage growth under control. This became a political flashpoint, and the administration’s focus on reducing immigration means that the equilibrium rate of new job growth is lower than in the past. This requires a shift in investor mindset when the closely watched non-farm payrolls report is released.
A recent study by the Dallas Fed noted that the pace of job growth needed to hold the unemployment rate constant could be as low as 30,000. This is much lower than 100,000 in late 2020 or around 250,000 in mid-2023. Said differently, a payroll report showing 30,000 new jobs would have been very disappointing in 2023, but now it is closer to equilibrium.
Messages from the Market
- Cyclical sectors have generally outperformed defensives
- Housing, Transports and some Consumer stocks have lagged
- Artificial Intelligence (AI) has been a key driver of markets
A popular way to think about markets and the economy is to look at which types of stocks are doing well and which are doing poorly. This was an approach made famous by legendary investor Stan Druckenmiller, who once quipped that “by far the best economic predictor I’ve ever met is the inside of the stock market”.
In the first four months of the year, defensive stocks significantly outperformed cyclical stocks. This coincided with increasing risks around trade tariffs, which peaked in early April. Equity markets bottomed in mid-April when tariff rates were paused and negotiations began.
Markets have staged a strong rally from the lows, with defensive sectors such as Consumer Staples and Health Care lagging.
Interestingly, a number of traditionally cyclical sectors have lagged. Transportation, housing, chemicals, and many consumer facing stocks have all underperformed the market. This suggests the market may not be giving the all-clear on the strength of the economy.
Understanding the economy and below the surface of the equity market has been complicated by a key theme now driving markets: Artificial Intelligence (AI)
The AI Factor
- Semiconductor stocks have been early beneficiaries of AI
- The AI buildout includes a number of different industries
Technology stocks performed very well in 2020 and 2021 when firms benefited from people staying at home and spending more time on their computers and mobile phones. That cycle became increasingly exuberant, with valuations for speculative technology stocks rising relentlessly. This exuberance came to an abrupt halt in 2022 as interest rates rose. A Goldman Sachs index of “Non-Profitable Technology stocks” fell almost 80% peak-to-trough.
Technological progress tends to come in waves, and we didn’t have to wait long for the next one. In late 2022, we saw the release of ChatGPT, OpenAI’s AI-powered chatbot. This was the catalyst for an investment super-cycle in the technology enabling the AI revolution.
Nvidia has been at the centre of this, as they make the cutting-edge semiconductors used for AI tasks. We have owned the stock since January 2018, and it has seen a remarkable rise, becoming the largest stock in the S&P 500 since General Motors in the 1970s.
While Nvidia is central to the AI theme, the story goes well beyond just one company. There are a number of key layers that span a range of sectors across both public equities and private markets:
The Scale of AI
- A small number of firms are making a huge bet on AI
- Technology firms do not want to be left behind
It is no exaggeration to describe the scale of investment into AI infrastructure as enormous. Large technology firms are projected to invest around $1.4tn between 2025 and 2027. For context, the total domestic investment across the entire US economy for 2024 was $5.3tn. A small number of mega cap companies are making a very large bet on the future of AI.
Former Intel CEO Andy Grove once remarked that “complacency breeds failure and only the paranoid survive”. Intel missed out on the smartphone revolution as it was generating considerable profits from the PC industry. When the market environment shifted, Intel was left behind and risked fading into irrelevance. Scale has been crucial in technology, with companies able to dominate in online search, social media, and online retail. There is a lot of uncertainty around the economics of AI, but with monopoly or oligopoly type profits potentially at stake, technology companies are investing out of necessity to be at the forefront of AI.
OpenAI CEO Sam Altman recently summed up this mentality, noting that “whether we burn $500 million a year or $5 billion or $50 billion a year, I don’t care. I genuinely don’t, as long as we can stay on a trajectory where eventually we create way more value for society than that and as long as we can figure out a way to pay the bills.
Accounting for AI
- The costs of AI are yet to show up in company earnings
When a company buys Nvidia chips, Nvidia recognises the revenue as soon as the goods are shipped. For the company buying the chips, this is capital investment which is capitalised on the balance sheet, for which they may also get a tax credit. While data centres are being built, semiconductors are classified as “under construction” and companies do not depreciate their assets until they come into use.
This creates a window where there is significant investment into AI, but the costs are not fully reflected in corporate profit and loss accounts. The key questions for the market are therefore how quickly these AI investments can be translated into revenue growth and what the return on capital will be.
Circulatory of AI Funding
- There are a lot of deals being done between AI companies
- This is not inherently bad, but does increase risk
A key factor in assessing the sustainability of AI spending comes from how it is being funded. Coming into the year, spending on AI infrastructure was predominantly from the large, highly profitable tech companies and funded from cash flow rather than borrowing. Some smaller companies have used leverage, but early access to cutting edge semiconductors gave them a valuable competitive advantage. More recently, we have seen an increase in borrowing, with technology companies issuing bonds in the investment grade bond market.
OpenAI remains a private company, but has recently announced a huge network of infrastructure deals totalling almost $1.5 trillion. This included a $300bn deal with Oracle, a $100bn investment from Nvidia, and strategic partnerships with AMD and Broadcom. For a company that is yet to turn a profit, to strike deals of this magnitude is certainly a warning sign for investors.
These deals are a form of vendor financing, which is not inherently bad, but does increase risk. For example, a manufacturer of expensive farming equipment allowing farmers to pay for the goods over time as harvests come in, makes sense. The key question is whether the payments can be serviced. If a farmer has a run of bad weather and harvests, they may not be able to pay for the equipment. For the manufacturer, not only do they have bad debts, but they will then be selling less products to farmers who are struggling.
The key is the cash flow of the customers. If companies pouring money into AI suffer from a “bad harvest” and cannot generate cash flow, this circulatory financing increases market risks.
Bubble Watch
- The AI buildout is very different to Dot Com
- But there are some signs that caution is warranted
- Return on capital is a key metric to watch
A key theme in almost every conversation at the moment is whether AI is a bubble. The general sentiment is that in recent months we have started seeing bubble like characteristics. The word “bubble” gets used far too often in the finance industry. Assets fluctuate and become cheap or expensive at various points in the cycle; this is typical market behaviour and different from a bubble. A bubble is generally defined as when assets rise far above their real, fundamental value, driven by speculation and herd behaviour.
In a bubble, prices disconnect from the underlying financials, which for equities means corporate profits. Using Nvidia as an example, this has not been the case. In 2023 Nivida traded at around 50x forward earnings, but is now closer to 35x. However, there is a question of whether the “bubble” is in earnings rather than valuation multiples.
The AI buildout certainly has echoes of past bubbles, but also notable differences. In the Telecom and Dot-Com bubble of the late 1990s the situation was “build it and they will come”. Significant amounts of fibre optic cable were laid, which took many years generate a profit, if at all. The AI buildout is fundamentally different. There is demand for computing power that cannot be met due to bottlenecks in data centre capacity and energy.
The optimistic case is that AI will not just generate profits from technology spending, but generate efficiencies throughout businesses across different sectors. For example, replacing people with AI in organisations could drive significant efficiencies, with technology companies capturing a share of the profits.
We have seen similar dynamics with cloud computing, social media/ online advertising and online retail. Cloud computing has allowed companies to take out costs of buying and maintaining hardware, and cloud companies have benefited. Social media and online retail companies have not only taken a share of existing advertising spending, they have allowed retail businesses to close physical shops and use the internet as a “shop front”. This has taken out costs and technology companies have again been able to capture profits.
Research by Goldman Sachs suggests that plausible estimates of the present-discounted value for the revenue unlocked by AI productivity gains in the US range from $5tn-$19tn. The future is inherently uncertain, but if such estimates prove to be correct then AI spending is justified. AI is already being used across multiple sectors. AI allows software development teams to run more efficiently, energy companies to drill more efficiently, while health care, law, and finance are also at the early stage of adoption.
It is difficult to break down AI investment as a share of the US economy. Technology investment has been a meaningful contributor to economic growth this year, but this also includes non-AI Technology. Goldman Sachs estimates that AI investment as a share of US GDP is less than 1%, which is much lower than prior large technology cycles which were closer to 2-5%.
The Large Language Models (LLM) are a key part of the AI story. Some of the leading models are private, such as OpenAI’s ChatGPT or Anthropic’s Claude, while Google owns Gemini and Meta owns LLaMA. China has also been making good progress. There is a debate around how many LLMs there will be, with this being a key determinant of the economics of AI. The question focusing on weather models will be commoditised and widely available at low cost, or will a small number of firms benefit from dominant market positions. Estimates at this stage from industry experts are that between 2-5 LLMs will become leading industry providers.
There are bubble-like dynamics in Venture Capital (VC), where start-ups are being funded with enthusiasm. VC is a very different market to public equities. There may be a “bubble” here, but the nature of that market is lots of businesses struggle or fail, while the winners see huge future profits. Having said that, weakness in VC could translate into lower demand for computing power, so this is a risk to watch.
Debasement and the Deficit Debate
- Gold has risen substantially this year
- Government spending levels have been in focus
- Inflation expectations have remained well behaved
The financial press has written a lot about the so-called “debasement trade” driving financial markets. The premise is that the value of money, particularly the US dollar, is being devalued against both physical assets and financial assets. The significant move higher in the gold price in the first nine months of the year was a key catalyst for this debasement argument.
With the US government pressuring the Federal Reserve to cut interest rates, concerns about debasement are understandable. However, the inconsistency here is that the definition of falling purchasing power (debasement) is inflation. Since August, market-based measures of expected inflation have been falling, not rising. Pass-through of trade tariffs to inflation have been lower than expected, as companies have managed their supply chains and avoided passing on the full cost increases to customers.
Another aspect to the “debasement trade” is the state of government finances. The US government spent a little over $7 trillion in the fiscal year 2025, while raising just $5.2 trillion in taxes. To cover the gap, the US borrowed $1.8 trillion, creating a budget deficit of around 6% of GDP. Given the projections for government debt levels in future decades, it is understandable to be worried about fiscal sustainability.
It has been an interesting year in bond markets, with government bond yields falling, not rising. The fall in bond yields (higher bond prices) has been more pronounced for short dated bonds compared to long dated bonds. Short dated bonds are more closely linked to interest rates. If central banks cut rates when cuts aren’t warranted, that is negative for the currency and leads to higher inflation i.e. debasement trade dynamics. The market is giving central banks the benefit of the doubt and that some cuts are warranted, but longer dated yields remaining more elevated does put governments on notice.
The US benefits from reserve currency status, whereas countries like the UK are more exposed to fiscal sustainability worries.
The US Dollar
- The dollar has weakened after a strong multi-year run
- Questions raised over the dollar’s safe haven status
- The bar to move money out of the US remains high
The dollar has been in focus this year. The dollar has been in a long bull market since the Global Financial Crisis of 2008, but questions have been raised as to whether the dollar has peaked. One of the attractive attributes of the dollar in recent years is that it has had both a high-interest rate compared to the rest of the world, and also appreciated in times of global stress. This is an unusual combination for any currency and made it an attractive allocation in global portfolios.
However, in the equity market drawdown of March and April this year, the dollar depreciated. This meant that unhedged overseas investors lost on both asset prices and the dollar. There have been worries that global investors would sell US assets, but instead what we have seen is global investors hedge a higher proportion of their dollar exposure.
Many investment portfolios, including a lot of our own, are heavily exposed to US assets. Our investment commentaries often tilt heavily towards the US, not because we don’t follow events in the Rest of World, but because what happens in the US is so fundamental to global financial markets. This is something that we are aware of and have been debating, but the bar to moving money out of the US while AI is outperforming is high. That said, overweighting Emerging Markets and Japan have both added value in our portfolios this year.
In Summary
- AI has been a key driver of equity markets
- Bonds have also contributed positively to performance
- Alternatives allocation have benefitted from exposure to gold
After equity weakness in the first four months of the year, it has been a very good period for multi-asset investment portfolios. AI and the associated data centre and energy build out have been key drivers. The US and global economy are far from firing on all cylinders. Parts of the US like housing and manufacturing are weak, while Europe, China, and the UK are muddling through a range of their own problems.
In equities, we have significant exposure to Technology and the AI theme more generally. The challenge for active managers has been that benchmark indices (e.g. S&P 500) also have significant exposure. Portfolios have benefitted from exposure to AI, but we are cognisant of the level of exposure. Selling Technology too early can be painful. Back in late 1996, then Fed Chair Greenspan noted that the market was showing signs of “irrational exuberance”. The market then continued to rally for another three years.
It is a cliché in the industry, but diversification has been very helpful for maintaining strong risk- adjusted returns. Bonds and equities were complementary for two decades until 2021, but have since become more correlated i.e. moving up and down together, which is less helpful for investment portfolios.
In fixed income allocations we have carefully managed risk, both in duration and especially in credit. We have a benchmark allocation to duration, but are more selective in credit, combining a high allocation to government bonds, with some smaller allocations to Emerging Market Debt and Short Duration High Yield. This has provided a nice blend of income, capital gains, but with managed downside risk.
Alternative allocations have been particularly beneficial this year. While we would not use the phrase “debasement trade”, our portfolios benefited from some of these themes. Our allocation to gold has been between 3.5 and 5.0%, which has been a strong contributor to performance. Our flagship fund of hedge funds is having a good year, driven by Long/Short Equity strategies, Credit strategies, and Macro funds. Commodities have been mixed, with oil prices muted due to heavy global supply, but strength in some of the industrial metals and agricultural commodities.
With the global economy muddling through, AI will continue to be a key theme in markets. This is very different to the Dot Com bubble, but the ability of the companies investing heavily in AI to generate an attractive return on capital is a key area of focus. If real world businesses can harness AI to drive efficiencies, that will be good for corporate profits, which drive the stock market. The dispersion between AI winners and AI losers is likely to grow, so differentiating between the two, while managing risk and portfolio correlations, is the focus as we head into 2026.