• Equity markets had a positive month as optimism rises for a soft economic landing
  • AI interest continues to fuel demand and innovation within a strong semiconductor sector
  • We are excited by the potential impact of AI to drive an inflection in productivity growth and create new markets as the next General Purpose Technology (GPT)


Market Review

Global equity markets continued to rally in March, the MSCI All Country World Net Total Return Index gaining +3.0% while the S&P 500 Index and the DJ Euro Stoxx 600 gained +3.2% and +4.1% respectively (all returns in sterling terms). This topped off a strong first calendar quarter for risk assets with several indices making new highs, including the S&P 500, DJ Euro Stoxx 600 and Nikkei 225 – finally surpassing its 1989 record high.

Equity indices were supported by economic data which suggests the global economy continues to deliver reasonable growth while the disinflation process remains broadly on track. Markets have moved higher this year on the back of data supportive of a soft landing (where inflation moderates without a severe increase in unemployment) and optimism around AI.

Job growth remains strong as the US economy added 275,000 jobs in February, above forecasts of 200,000, with the largest employment gains occurring in healthcare and government sectors. Average hourly earnings rose +4.3% year-on-year (y/y), below forecasts of +4.4%. After a hotter January print, the US Consumer Price Index (CPI) annual inflation rate edged up to +3.2% year on year (y/y) in February, above forecasts of +3.1% y/y, but the Federal Reserve’s (Fed) preferred measure, core Personal Consumption Expenditure (PCE), came in slightly below consensus at +2.8% y/y. Brent Crude (+4.6% during the month) rose back above $85 for the first time since October on escalating geopolitical tensions in the Middle East.

As widely anticipated, the Federal Open Market Committee (FOMC) unanimously voted to leave the Fed funds rate unchanged at a 23-year high of 5.25-5.5% for a fifth consecutive meeting in March. The FOMC statement reiterated "the Committee does not expect it will be appropriate to reduce the target range until it has gained greater confidence that inflation is moving sustainably toward 2% ", but still forecast three rate cuts before the end of 2024 (unchanged from December) despite modestly higher GDP and inflation estimates.

Technology review

The technology sector slightly lagged the broader market in March, the Dow Jones Global Technology Net Total Return Index (W1TECN) gaining +2.7%. Large-cap technology stocks lagged their small and mid-cap peers; the Russell 1000 Technology Index (large cap) and Russell 2000 Technology Index (small cap) gaining +2.2% and +4.2% respectively. Given the pace of the AI infrastructure buildout, the Philadelphia Semiconductor Index (SOX) continued to lead, returning +3.8% (supported by NVIDIA +14%), while the NASDAQ Internet Index and Bloomberg Americas Software Index returned +1.7% and +0.3% respectively.

There were some notable results during the tail end of earnings season. In the semiconductor sector, Broadcom’s report was broadly in line with forecasts. While management did not raise full-year guidance, the networking segment was very strong, driven by an inflection in AI demand, with AI-related revenue growing +50% (quarter on quarter (q/q) up 4x y/y) to $2.3bn, offsetting weakness in other segments. Recently acquired VMware had a strong start and is expected to grow sequentially throughout the year. Marvell Technology Group reported in-line results, but next quarter guidance was soft due to weak demand outside AI, particularly from telecommunications carriers and enterprise networking customers. While we have reduced our position, the data centre business grew +54% y/y, driven by the strength of AI demand, and management expects further growth as the company’s AI silicon programs start to ramp up.

The technology sector slightly lagged the broader market in March…[though] there were some notable results during the tail end of earnings season.

Memory supplier Micron Technology’s results and guidance were well ahead of expectations, driven by both AI and non-AI products. Management expects improving demand for memory, combined with supply constraints, to drive prices higher in 2024 and 2025, leading to record revenue and profitability going forward. Semiconductor production equipment vendor Disco also gave positive commentary about demand for its equipment for silicon carbide and high-bandwidth memory markets (for AI applications), and benefitted from the announcement of further stimulus from the Japanese government for semiconductor projects that favour local suppliers.

The other major event in the semiconductor industry was NVIDIA’s GTC (GPU Technology Conference), at which CEO Jensen Huang launched the Blackwell graphics processing unit (GPU), including the B100, B200 and GB200 superchips, as well as upgraded switch offerings and a host of partnership announcements. The performance metrics were impressive, particularly for the GB200, where NVIDIA is vertically integrating to deliver a c75% reduction in power consumption for large language model (LLM) training. Quanta Computer rallied after it emerged as one of the leading server manufacturers (at the rack level) for NVIDIA's new GB200 NVL AI server offering.

Advanced Micro Devices (AMD) underperformed due to the excitement surrounding NVIDIA’s new products. However, hyperscale cloud customers who are deploying AMD's MI300X GPU are seeing strong performance across training and inference and supply chain data points appear to support strong growth for AMD ahead (and likely improved sentiment at some stage, when they discuss their own product roadmap in more detail). ARM Holdings (ARM) also lagged after a period of significant outperformance in February. The IPO lockup expired on 12 March exerting downward pressure on the stock price, even if Softbank, ARM's largest shareholder, did not monetise any of their stake.

In the software sector, cybersecurity vendor CrowdStrike Holdings reported solid results and guidance. Annual recurring revenue (ARR) was well ahead of expectations. Management also highlighted a large cloud security deal with a “very fast-growing AI company”, providing them with a full suite of products that can protect LLM development.

Search and analytics platform Elastic reported results and next quarter guidance modestly above consensus forecasts but light of buy side expectations after a strong run, with continued softness in the small and medium-sized business market offset by strength in the enterprise market. MongoDB also missed expectations with lower than expected full-year revenue growth guidance. Underlying customer growth and consumption trends appear to be improving but management is accelerating investment, having dialled back last year, impacting operating margins in the near term. Adobe Systems (underweight (u/w)) delivered less net new ARR growth upside than in previous quarters and issued guidance below consensus estimates, failing to quell concerns that the risks posed by AI might outweigh the opportunities. More broadly, AI continues to cast something of a shadow over the software sector as investors grapple with what risk is posed to existing companies should AI materially change the architecture of future software applications. For instance, Klarna* recently announced a custom-developed AI assistant which after just one month is said to be doing the work of 700 full-time call-centre agents. The company estimates this could drive $40m in profit improvement this year. A supportive datapoint for AI adoption, but less obviously good news for existing packaged software vendors.

In the internet sector, Alphabet recovered from its underperformance in February, benefiting from press reports suggesting Apple are in discussions to license Alphabet’s Gemini AI models for upcoming iPhone software features this year. Despite this, Apple’s stock price was weak due to regulatory headwinds including a new Department of Justice antitrust lawsuit during the month, concerns about demand and competition in China and uncertainty about the company’s AI strategy.

Outlook

We believe generative AI represents a rare example of ‘discontinuous’ technology progress as performance/capability improves at a supernormal rate. This is akin to how the Bessemer process improved the price/performance of steel to such an extent that the New York skyline was transformed by the skyscrapers it enabled over the following half-century. Sitting here now, it is hard for most of us to imagine just how significantly generative AI is likely to reshape the landscape in almost every industry over the next decade, let alone predict the likely significant new markets it creates.

We remain incredibly excited by the potential impact of AI to create new markets and drive an inflection in productivity growth.

We have been technology investors for over 25 years and the pace of innovation since ChatGPT was launched in November 2022 is by far the fastest we have experienced. We remain incredibly excited by the potential impact of AI to create new markets and drive an inflection in productivity growth. While there are some parallels with the dotcom period, it feels more like 1995-96 – the beginning, rather than the end of a transformative period.

As is typical in new cycles, generative AI brings with it a major new computing architecture shift (in this case from serial to parallel computing), with virtually every single component of computing infrastructure reimagined. This stands in contrast with the ‘general purpose’ compute which has been the bedrock of cloud computing architecture, where the flexibility of the infrastructure (both in terms of elastic scaling and multiple different applications running on common hardware) is at the core of the value proposition. This move from general purpose compute to accelerated compute necessitates an entirely new supply chain and each individual component of an AI server has to be completely redesigned for an AI world.

We have long believed a growth-centric approach is best within the technology sector, but particularly so at times of accelerated technology disruption (winners from a new technology outgrow losers). As with prior discontinuous technology changes and computing architectural shifts, we believe there will be strong growth ahead but significant challenges for some incumbents (the so-called ‘incumbent’s dilemma’). Already, mega-cap returns are bifurcating with the so-called ‘Magnificent Seven’ rebranded as the ‘Fab Four’ which fits with the Trust’s significantly reduced position in Apple, less Alphabet and a tail position in Tesla, while maintaining larger positions in NVIDIA, Microsoft, Meta Platforms (Facebook) and Amazon.

Our reduced Apple holding reflects several regulatory headwinds and increasing competition in China, as well as our view that the company is currently behind in terms of AI. We will look for meaningful new AI announcements (perhaps around its developer conference in June) to rebuild our position. We also remain underweight Alphabet as generative AI could threaten the company’s dominant position (and/or economics) in search, despite its strong AI technical capabilities, talent base and data assets. While it is too early to label these technology behemoths as generative AI losers, they both have significant incumbency challenges to navigate and we will watch their progress closely.

We have also made several other noteworthy portfolio changes over the past year. First, we reduced our direct Chinese holdings significantly because of the ailing economy, increased geopolitical tension and export controls. Second, we cut electric vehicle/automotive-related exposure due to macroeconomic and competitive pressures; we have a minimum position in Tesla and negligible exposure to automotive semiconductors for now. The same applies to energy transition (for us largely solar-related stocks), which we have exited because of inventory issues, macro headwinds and higher borrowing costs (interest rates).

These changes have allowed us to continue to broaden and deepen our exposure to AI which we believe will make or break investor performance over the next decade, not just within technology but eventually across most sectors as history has demonstrated with earlier general purpose technologies.

More than six years ago, we launched a dedicated AI strategy within the technology team at Polar Capital. This has allowed us to accumulate knowledge and experience in a space that is broad, and rapidly evolving. To date, this has enabled us to deepen and diversify PCT’s exposure to the AI hardware opportunity, beyond understandably well-owned US chipmakers such as AMD and NVIDIA. Our team of 10 dedicated fund managers and analysts are travelling extensively, meeting company management and industry experts regularly. We are also using generative AI tools ourselves, including innovative AI-infused packaged software (such as Quartr and AlphaSense) and several AI projects of our own to enhance our analyst productivity and our screening processes. The more we use AI and the more we meet with and listen to the visionary founders of companies such as OpenAI, Scale AI, Anthropic and Perplexity AI, the greater our conviction has grown.

We therefore strongly believe the market is still significantly underestimating the true potential of generative AI. Even within the technology sector, however, this has led to a noticeable divergence in returns so far this year. For now, we expect these trends to continue to support the Trust’s positioning and we remain constructive and fully invested. That said, as with previous cycles, rapid change brings uncertainty; investors should be prepared for and able to tolerate sentiment swings (both positive and negative) and bouts of increased volatility during this exciting transition period.

*not held