• Positive macro developments: first Federal Reserve rate cut and Chinese monetary stimulus
• Market volatility is to be expected given we are in the early stages of an AI technology cycle
• Recent datapoints from OpenAI and Meta support our view that AI adoption will gather pace
  

Market review

After a volatile August, global equity markets delivered positive returns in September, pushing indices back towards year-to-date highs. However, this was largely offset by sterling strength (+2.1%) leaving the MSCI All Country World Net Total Return Index largely unchanged (+0.2%) during the month, all figures in sterling terms.

The strong market performance was supported by the start of the rate-cutting cycle in the US and the biggest month of global monetary easing since April 2020. The Federal Reserve (Fed) delivered a 50 basis point¹ (bps) cut at its September meeting despite many investors expecting 25bps. The more dovish² 50bps combined with positive economic commentary to leave markets relatively comfortable that monetary policy is not too restrictive and that the Fed will act to mitigate further labour market weakness if required.

In addition, economic data during the month indicated a solid underlying economy. The weekly initial jobless claims four-week moving average fell from 241,000 in early August to 225,000 in September, the lowest level since May. Meanwhile, September’s employment numbers indicated 142,000 workers were added in August, while the unemployment rate dropped to 4.2%. The supply of labour appears to be still growing at a slightly higher rate than labour demand, which puts upward pressure on the unemployment rate, even without an inflection in layoffs.

Oil prices fell during the month potentially underpinning a softer inflation outlook. This was despite a further escalation of political tensions in the Middle East as Israel intensified its efforts against Hezbollah in Lebanon and Iran launched ballistic missiles in response. While it is hard to predict what happens next, an Israeli response is expected; if it targeted energy infrastructure (either oil supply or a setback of nuclear ambitions), energy prices could rise. If there is a positive, the US economy is less sensitive to energy prices than historically.

China and China-exposed equities performed very strongly into the end of the month following announcements of significant government stimulus measures. These involved both monetary and fiscal policy support to help achieve 5% GDP growth for the year and included a host of measures to boost the property market. In addition, the central bank cut mortgage rates by 50bps as the focus shifts to stimulating consumer spending. The Shanghai Shenzhen CSI 300 Index rose +27% from 13 September, rebounding from a five-year low to a one-year high.

China and China-exposed equities performed very strongly into the end of the month following announcements of significant government stimulus measures.

Strength in China pressured other Asian markets as money rotated within the region. Our perspective is that the stimulus is welcome given property, demographic and trade/tariff-related headwinds to the world’s second-largest economy. However, ahead of a close-run US presidential election, we remain sensitive to the risk of higher tariffs should Donald Trump prevail. We also cannot know how successful the measures will ultimately prove given the structural issues that remain. Nonetheless, the stimulus should be supportive for the global economy and provide some of our stocks with welcome cyclical tailwinds.

Technology review

The technology sector outperformed the broader market in September as the Dow Jones Global Technology Net Total Return Index (W1TECN) returned +0.5%. Large-cap technology stocks outperformed their small and mid-cap peers; the Russell 1000 Technology Index (large cap) and Russell 2000 Technology Index (small cap) returned 0.7% and -2.1% respectively. The Philadelphia Semiconductor Index (SOX) fell 1.7% while the NASDAQ Internet Index and Bloomberg Americas Software Index returned +3% and +1.6% respectively (all figures in sterling terms).

There were a few notable off-season earnings reports during the month. In the semiconductor sector, Broadcom delivered a mixed quarter, with strength in VMware – its cloud computing and virtualisation business – offset by softer than expected results for the semiconductor business where it is seeing a slower recovery in cyclical non-AI end markets. AI semiconductor sales were only flat quarter-on-quarter (q/q), but management had messaged that orders would be lumpy, and they guided for +10% q/q growth next quarter and raised 2024 guidance to >$12bn (up from $11bn).

Broader economic fears and lacklustre PC, smartphone and automotive demand also weighed on other semiconductor stocks. In hardware, component supplier Amphenol lagged during the month due to uncertainty about NVIDIA’s product ramp, as well as automotive end market weakness. Concerns about memory and HBM (high bandwidth memory) spending in 2025 and movement in the yen had a negative impact on Disco which sells semiconductor manufacturing equipment and precision tools used to produce HBM.

Smartphone battery supplier TDK was caught up in Japanese market weakness at the end of the month, after Japan’s governing party chose Shigeru Ishiba as its next prime minister; he has been a critic of the country’s longstanding ultralow interest rates. KLA, which designs and manufactures yield management and process monitoring systems for the semiconductor industry, was negatively impacted by concerns about wafer fab equipment (WFE) spending in 2025, due to slowing capital expenditure by Intel* and Samsung as well as the potential for tighter restrictions on trade with China.

Memory provider Micron Technology bucked the trend with stronger than expected results, with revenue +93% year-on-year (y/y). Data centre demand remained robust, driven by the AI infrastructure buildout, which offset weakness in non-AI markets. Gross margins were also robust, benefiting from an improved product mix and higher than expected HBM production cost reductions. Positive trends are expected to continue with next quarter guidance for revenue +12% q/q, above market forecasts. Advanced Micro Devices (AMD) also performed strongly during the month, as the company stands to take further server market share from struggling rival Intel* in the core CPU business. Micron Technology’s strong results and guidance, along with reassuring commentary regarding the ramp up of flagship Blackwell AI chips in Q4 by NVIDIA CEO Jensen Huang at the Goldman Sachs Technology Conference in San Francisco (attended by team members Nick Evans and Lina Ghayor) prompted a sharp recovery in AI-exposed semiconductor companies. SOX rose +12% from 6 September.

In software, Oracle reported solid results and guidance, reiterating full-year guidance for double-digit revenue growth. The company’s backlog growth increased to 53% y/y and reached $99bn, driven by strong AI-related demand; this emboldened management to raise its FY26 revenue target to $66bn and initiate an ambitious FY29 revenue target of $104bn. It expects capital expenditure to double this year as it invests in infrastructure to support this growth, which is another positive data point for NVIDIA and other AI semiconductor and infrastructure providers.

Adobe Systems* reported strong quarterly results but the stock was weak due to a cautious Q4 outlook, while management provided only limited incremental detail on AI monetisation. Despite a modest rebound this month, we believe the outlook for the broader software sector remains challenging, due in part to uncertainty about the role of generative AI in reshaping the application landscape and the budget reallocation to non-software AI initiatives.

Arista Networks also rallied during the month, driven by encouraging commentary around its hyperscale AI networking trials and winning a fifth pilot for AI backend networking with ethernet. We initiated a position in Ciena, a provider of networking solutions. The company reported solid results above consensus expectations, although next-quarter guidance was slightly below. It is poised to benefit from rising capital investment from cloud service providers as a result of AI-driven traffic growth. Tesla also rallied during the month on strong car registration data from China and excitement about the upcoming ‘Robotaxi’ event.

Outlook

Financial conditions have loosened significantly in the past few months. The Fed’s 50bps cut is a further signal that it is willing to act decisively if it feels the risks around further labour market weakness significantly outweigh those around a resurgence in inflation. This should provide a supportive backdrop for growth technology companies as economic growth moderates, but more importantly a severe recession is likely avoided due to pre-emptive policy action.

There are likely to be regular bouts of market volatility, however, given both the backdrop of elevated political risk and the limited data points (confirmatory or dissenting) regarding the progress of AI, which is a normal feature of early technology adoption cycles. For example, between 1995 and 1998 – the internet years prior to the dot.com ‘melt up’ – there were nine NASDAQ Index corrections of 10% or more (seven of which were of -15% or greater). Over this volatile period, the Index rose by c350% (all returns in USD terms).

Many recent AI data points are supportive of our view that we are in the early stages of the next general purpose technology cycle... AI represents a rare example of discontinuous technological change.

Many recent AI data points are supportive of our view that we are in the early stages of the next general purpose technology cycle. The first nationally representative survey of generative AI adoption indicated that in August 2024 39% of the US population aged 18-64 used generative AI. Adoption appears to be faster than previous technology cycles with AI adoption at 39% less than two years after ChatGPT’s launch in November 2022, compared to 20% for the internet and PC after two and three years, respectively. Recent news suggests that adoption might even be accelerating – ChatGPT is said to have reached 250 million weekly active users (WAU), up from 200 million in August, while Meta Platforms recently increased its AI monthly active user (MAU) number to “nearly 500 million” from the 400 million disclosed in August. Rapid growth likely reflects AI’s broad applicability and low barriers to consumer experimentation (a smartphone and internet connection). However, it also supports our long-held view that AI represents a rare example of discontinuous technological change.

We have previously said that the pace of innovation around generative AI is like nothing we have seen in our experience of investing across many technology cycles, since the late 1990s. Recent weeks have only cemented this view, through a combination of product announcements, AI-related growth/engagement statistics and our meetings with companies at investor and industry conferences. This month, OpenAI launched its latest o1 model, which includes advanced reasoning capabilities and has shown significant improvement in specific tasks using a ‘chain of thought’ process. Encouragingly, OpenAI also recently raised $6.6bn in new funding at a $157bn valuation; it is said to be on track to deliver revenue of $3.7bn in FY24 and anticipating $11.6bn in FY25.

There have been several other positive updates on the early monetisation of AI. NVIDIA’s CEO reiterated his claim that customers can generate $5 in rental revenue for $1 spent on NVIDIA infrastructure. Microsoft’s CTO Kevin Scott was clear that “we are demonstrably not at the point of diminishing marginal returns on how capable these AI systems can get”. FinTech provider Stripe* also reported that AI startups are growing at a significantly faster rate than the software-as-service (SaaS³) companies that came before them. The 100 highest-revenue AI startups on its platform took a median of 20 months to reach $30m+ in annualised revenue, five times faster than for the equivalent SaaS companies during the SaaS boom in 2018. There has also been interesting early data on the cost and efficiency benefits of adopting AI, with a report from Bain indicating that early generative AI initiatives could be worth up to 20% of EBITDA.

These (very) early adoption, technological progress and monetisation signals underpin our conviction in the AI theme and support the willingness to invest aggressively in the AI infrastructure to support this technological revolution. Blackstone’s CEO Stephen Schwarzman spoke to the “explosive trend” requiring $1trn of capital expenditures in the US over the next five years to build new data centres, and another $1trn outside the US. AI is already a key driver of technology buying decisions, up significantly from 13% in 2023, according to the Telarus Tech Trends Report.

Generative AI may only be c3% of IT budgets today according to a recent Goldman Sachs CIO survey, but it is expected to grow threefold over three years to c9%. This was the approximate level of cloud enterprise workload penetration in 2016 at which point Intel’s* relative underperformance versus the semiconductor sector began in earnest, so we remain vigilant about avoiding companies we believe may ultimately prove to be on the wrong side of AI. As a reminder, McKinsey believes generative AI could automate 30-50% of tasks in c60% of occupations by 2030.

Another way of thinking about the scale of the AI opportunity is to consider that today $5trn is spent on IT compared to $40trn on the wages of knowledge workers. Should AI begin to substitute, rather than augment, labour, the total addressable market (TAM) could expand from 3-9% of IT budgets to as much as 20-50% of total operating expenditure. For example, the contact centre software market is today estimated at $23bn to support 17 million human agents. However, as AI begins to replace humans, the TAM could exceed $500bn considering the fully loaded cost of an agent is likely $30-40kpa. This explains why ‘buy now/pay later’ leader Klarna has cut its headcount from 5,000 to 3,800 with suggested annual savings of $40m pa. Klarna has also disclosed that it has used internally developed AI tools to replace both Salesforce.com and Workday (neither held), supporting our concerns about the risks posed to enterprise software by AI.

As such, we remain extremely positive on the outlook for AI and believe 2025 will see others increasingly share this view as AI becomes more pervasive. However, elevated volatility may continue to affect markets, especially during October given its empirically unfavourable seasonality. This is typically worse in presidential election years, but the subsequent seasonal rebound in Q4 is also typically stronger. We are also mindful of heightened sensitivity to labour market/consumer spending data given some recession concerns. We are also entering third quarter preannouncement season in which news flow tends to be negative. Geopolitical risk also remains elevated, with a close-run presidential election just a month away.

While caution is understandable, we expect it to subside in the coming months and allow the market to climb the proverbial ‘wall of worry’. As we enter 2025, we also expect strong demand for NVIDIA’s Blackwell (and alternatives from competitors such as AMD and Broadcom) as well as high-profile generative AI model upgrades/product releases to reignite interest in AI-exposed stocks. It will also become more obvious that AI represents a rare moment of discontinuous technology progress that we expect to reshape most industries. While companies that successfully embrace AI may emerge as the new disruptors, like earlier GPTs AI also represents an existential threat to those who fail to grasp its significance.

* not held

1. A basis point is a common unit of measure for interest rates and other percentages in finance. Basis points are typically expressed with the abbreviations bp.

2. (Opposite of hawkish) Expansionary monetary policy; more likely to see lower interest rates, central banks buying government bonds and lowering the reserve requirements for banks.

3. SaaS allows users to connect to and use cloud-based apps over the internet.