• February saw resilient markets despite higher-than-expected inflation and an upcoming US presidential election
  • A positive month for the technology sector as AI demand remained strong
  • Continuing signs of generative AI’s productivity efficiency further support our conviction that it should offer a long-term secular tailwind

Market Review

Global equity markets continued to rally in February, the MSCI All Country World Net Total Return Index gaining +5.1%, while the S&P 500 and the DJ Euro Stoxx 600 indices returned +6.2% and +2.3% respectively (all returns are in sterling terms). The S&P 500 has delivered positive returns for 16 of the past 18 weeks, a run last seen in 1971.

Equity markets remained firm despite the challenges at New York Community Bancorp due to commercial real estate losses (which do not appear to be systemic), a higher-than-expected consumer price inflation print and uncertainty about the upcoming US election. This proved insufficient to derail the prevailing bullish narrative that the disinflation process remains on track with a soft landing (where inflation moderates without a severe increase in unemployment) a likely outcome, while the economy may be entering a period of higher productivity growth.

The US economy added 353,000 jobs in January, above forecasts of 180,000 and up from an average of 255,000 jobs per month in 2023, signalling that the labour market remains tight. The largest employment gains occurred in professional and business services, health care, retail trade and social assistance, with strength broader than recent months. Average hourly earnings rose +4.5% year-on-year (y/y) which helped push the 10-year Treasury yield higher by around 35 basis points to 4.25% during the month.

The US Consumer Price Index (CPI) annual inflation rate was +3.1% year-on-year (y/y), decelerating from +3.4% y/y in December but higher than forecasts of +2.9% y/y. Core CPI, which excludes volatile items such as food and energy, was steady at +3.9% y/y, a two-and-a-half year low. Personal Consumer Expenditure (PCE; the Fed’s preferred measure) was in line with expectations against fears of a strong print, but still increased to +2.5% from below 2% in December on a six-month annualised basis.

The minutes from the January Federal Open Market Committee (FOMC) meeting reiterated that Federal Reserve (Fed) officials see significant progress in bringing down inflation, but most “noted the risks of moving too quickly to ease the stance of policy” and “emphasised the importance of carefully assessing incoming data in judging whether inflation is moving down sustainably to 2%.” More confirmation that disinflation is becoming entrenched will be required before the Fed cuts rates, or prepare the ground for a decision in May to slow the pace of quantitative tightening (QT).

Geopolitical risks remain elevated as Houthi rebels continued to attack ships in the Red Sea; Brent crude oil prices increased a further +2% during the month. More encouragingly though, our view of moderating inflation and solid economic growth remains intact which ought to be supportive for the technology sector outlook. Our recent meetings with company management teams suggest this stable IT spending outlook with strong AI demand tailwinds remains in place.

Technology Review

The technology sector outperformed the broader market in February; the Dow Jones Global Technology Net Total Return Index rising +7.2%. Much of the relative outperformance of the sector was driven by the robust performance of AI-exposed companies, due to almost universally strong data points on demand/adoption and future capex during recent results.

Encouragingly, participation has broadened well beyond bellwether NVIDIA (which is now approaching 10% of the Dow Jones Global Technology Index and therefore also hitting UCITS limits), which has provided a welcome opportunity to add value through stock selection. NVIDIA’s strong performance underpinned robust absolute sector performance but large-cap technology stocks  were outperformed by their small and mid-cap peers; the Russell 1000 Technology Index (large cap) and the Russell 2000 Technology Index (small cap) returning +7.6% and +9.8% respectively.

Our recent meetings with company management teams suggest this stable IT spending outlook with strong AI demand tailwinds remains in place.The Trust’s relative performance also benefited from its underweight positions in Apple (u/w), which is being impacted by competition in China, app store regulation and an uncertain generative AI (GenAI) strategy. We have reduced the Trust’s Apple position substantially for now. Microsoft (u/w) also lagged slightly, pausing for breath after a strong run, despite delivering solid results. This was partially offset by Meta Platforms (Facebook) (u/w) which delivered a strong earnings report with improved advertising spending in key segments (e-commerce, consumer packaged goods and gaming) supported by tailwinds from earlier AI investments.

Given the pace of the AI infrastructure buildout, the Philadelphia Semiconductor Index (SOX) led, returning +11.9%, while the NASDAQ Internet Index and Bloomberg Americas Software Index returned +7.3% and +4.1% respectively.

NVIDIA delivered another standout quarter and issued guidance well above market forecasts. The data centre segment grew +27% quarter-on-quarter (q/q) and +408% y/y in the company’s fourth fiscal quarter (F4Q) driven by GenAI-related demand for the company’s graphics processing units (GPUs), with strong sales to large cloud service providers in the West. It is remarkable that NVIDIA were able to beat expectations comfortably, while China sales “declined significantly” due to US government restrictions. It is also notable that this was before the ramp of many of their new products and was achieved while Taiwan Semiconductor Manufacturing Company (TSMC) capacity remains tight. CEO Jensen Huang sounded confident about continued growth in “CY25 and beyond” allaying fears of a peak or plateau from high levels, based on expanding customer end markets (not just cloud), sovereign AI demand (as countries look to build their own large language models (LLMs) and encourage domestic innovation). Strength in inference (as well as training), which accounted for 40% of data centre revenue in the quarter addressed bears who believe NVIDIA is best suited to training only. Management also noted that next-generation products are essentially sold out as demand far exceeds supply, despite TSMC ramping additional capacity into 2H24.

ARM Holdings (ARM), a leading semiconductor IP company, also performed strongly on better-than-expected results, with both licensing and royalty revenue well ahead of market forecasts. The company also issued positive guidance, benefiting from increased penetration of its new ARM v9 architecture, which has a royalty rate double that of ARM v8, on average, and significantly more for Data centre chips used for AI applications.

Strong results from ARM and NVIDIA spurred a rally in other names exposed to the AI infrastructure buildout, including Advanced Micro Devices (AMD). There were some less favourable reports in the subsector, however. Chip and silicon IP solutions provider Rambus reported an in-line quarter, but next quarter guidance was below expectations. We exited our position because we do not believe their content gains in AI servers are enough to offset cannibalisation impact on legacy servers where demand remains soft. Quanta Computer also lagged during the month due to concerns about trends in the general (non-AI) server market, but here we believe AI price/content gains should be sufficient to drive future growth.

Enterprise storage company Pure Storage reported solid results and forward guidance for 10% revenue growth and a 17% operating margin was better than feared. The company also signed an eight-figure deal with a “major GPU cloud provider” and expressed confidence in future growth because price/performance vs hard disk drive has improved markedly aided by new E-series product introductions.

In the internet subsector, e-commerce platform Shopify delivered strong results, with gross merchandise value (GMV) +23% y/y, and revenue and operating profit ahead of expectations. Next quarter revenue guidance was also ahead, but operating expense guidance was higher than expected as management opted to increase online and offline marketing investments. Online advertising platform, Trade Desk, announced better than feared results and forward guidance, benefiting from secular shifts to retail media (ads that are featured on a retail website, app or marketplace) and connected TV. Commentary was encouraging regarding upcoming product launches, and management believes the company is well positioned to weather Alphabet’s third-party cookie deprecation this year, benefiting from the continued adoption of its alternative identifier, Unified ID 2.0.

Uber Technologies (Uber) reported gross bookings accelerating 1ppt to +21% y/y, benefiting from traction in new verticals like grocery, the Uber One subscription programme and advertising. At a subsequent investor update, the company issued new long-term guidance with 2026 gross bookings (growing mid-high teens for the next three years) and EBITDA ahead of expectations. They also announced a $7bn buyback and reduced share-based compensation. Food delivery platform, DoorDash, reported results above market forecasts, with gross order value +22% y/y (ahead of Uber Eats at +19% y/y) although the magnitude of EBITDA upside was less than previous quarters. 2024 guidance for gross order value to grow +15% was below elevated buyside expectations but appears conservative given Uber’s more bullish guidance.

Much of the relative outperformance of the sector was driven by the robust performance of AI-exposed companiesResults were mixed in the software subsector. Datadog, a monitoring and analytics platform for large-scale applications and infrastructure, reported full year revenue growth (+21% y/y) and operating margin guidance modestly below estimates. This could be conservative given usage by some of Datadog’s largest customers is growing and the contribution from AI increased again in 4Q23 (to 3% of annual recurring revenue, up from 2.5% in 3Q23), which could be an upside driver for 2024.

Cloud services provider CloudFlare, however, rallied after a strong print driven by federal and education verticals, as well as large enterprises, while full year guidance was better than expected. Management reiterated the company’s strong positioning in GenAI, particularly around inference-related tasks. CloudFlare now has inference-optimised graphics processing units (GPUs) running in >120 cities and should have coverage in nearly every location in which the company operates by the end of 2024.

Monday.com reported revenue growth of +35% y/y but missed elevated expectations after positive channel checks had driven a strong move up in the stock ahead of the print. FY24 guidance was at the low end of the FY24-26 guidance range for ‘high-20s to low-30s’ revenue growth shared at the recent analyst day. This may prove to be conservative, however, given that a c20% price increase should start layering in from this month. Customer relationship management platform, HubSpot, reported inline results given that SMB spending has been soft, with more “urgency” seen from larger customers in December. Management do not yet see a demand inflection on the horizon, however, prudently forecasting +18% revenue growth in 2024. Encouragingly, early feedback suggests HubSpot’s AI solutions are making a difference, particularly for their SMB customers where it can significantly improve the productivity of small marketing teams.


We have long been excited about the potential for AI to change the technology landscape. More than six years ago we launched an adjacent AI strategy to capture opportunities within, as well as beyond, the technology market. At the time, we were certain the AI opportunity was huge but it was unclear when the technology would reach the so-called ‘tipping point’. It now appears the release of ChatGPT in November 2022 was that moment. Even as “AI maximalists” we have been positively surprised by the current pace of AI innovation and now expect a more rapid timeline to disruption of non-technology sectors. We also believe the experience gained since we launched our AI strategy has allowed us to pivot our core technology funds (including the Polar Capital Technology Trust) quickly to benefit from the expected acceleration in AI demand which we believe is just getting underway.

Investor opinions are clearly divided on the impact of AI (even among technology fund managers) with many sceptical about the sustainability of the trend; this bifurcation of positioning is becoming visible in performance too. We believe this is the start of a major new cycle and not the end. The abundance of positive AI news flow, new product/model releases and strongly supportive data points during fourth-quarter earnings season is hard to argue with. GenAI is a ground-breaking technology that, while in its infancy, is already incredibly powerful. It is easy to pick on the failures (including the high-profile release and subsequent withdrawal of Google’s Gemini image generation features) and/or undoubted regulatory risks (manageable in our view), but it is hard to deny the pace of innovation in the past 18 months has been anything other than incredible.

OpenAI released GPT-4 in March 2023, just four months after GPT-3.5, and is significantly more capable across a wide range of tasks. GPT-4 Turbo followed at the September Developer Day alongside multimodal capabilities including vision and text-to-speech. The extraordinary text-to-video Sora tool was announced just this month. Competition is intense: Google has also released multiple Gemini models in different sizes for different use cases and underlying chips, and startups such as PerplexityAI, Anthropic (Claude 3) and Mistral (Le Chat) are already in public use and comparable in performance to both OpenAI and Google on some metrics. Meta Platforms (Facebook) has released a range of open-source models which have seen widespread adoption including Llama 2 and an ecosystem of data and model tooling has sprung up including ScaleAI (data preparation), LangChain (app development) and HuggingFace (model repository).

A number of team members have attended technology conferences during the past few weeks. Nick was recently in San Francisco, meeting with many technology company CEOs and has said he cannot recall a time when a new technology has moved from the hype phase to widespread adoption so quickly. This is not a product cycle like smartphones or 5G, or a distant concept like the Metaverse. Instead, GenAI looks like the next general-purpose technology (GPT) in the making.

This appears to be supported by strength in hyperscale capex, as the four hyperscalers (Amazon, Google, Meta Platforms (Facebook) and Microsoft) raised aggregate 2024 capex growth expectations from +18% to +26%, with the incremental spend focused on AI. These are the early adopters, and they are seeing what AI is capable of within their own businesses. Corporates and governments alike will be compelled to invest in AI in search of productivity gains and to remain competitive: NVIDIA spoke to strong demand from sovereign buyers in their most recent quarter. We are also seeing early evidence of corporate buyers preparing for AI adoption which could presage an AI-ready hardware upgrade cycle in the coming quarters, even as near-term enterprise spending trends have been lacklustre.

One example is Klarna*, who announced last week they had built an AI assistant with OpenAI which, in its first month, has taken over two-thirds of Klarna’s customer service chats and is doing the equivalent work of 700 agents with a 25% drop in repeat enquiries and resolution time of two minutes versus 11 minutes previously. Klarna believe this will drive a $40m improvement in profit this year.

We have been positively surprised by the current pace of AI innovation and now expect a more rapid timeline to disruption of non-technology sectorsAt our recent Polar Capital Investor Conference in London we likened the difference between general-purpose compute (serial, cost efficiency driven, flexible) and accelerated AI compute (parallel, performance-driven, specialised) to the difference between a Toyota Prius and a McLaren Formula 1 car: every single part needs to be rethought to solve for a different use case. This shift presents myriad active investment opportunities among the foundational technologies and the ‘new AI stack’ supply chain. As Anthropic CEO, Dario Amodei, put it: “Whenever you do something at a scale where it's never been done before, every single component, every single thing has to be done in a new way”. NVIDIA CEO Jensen Huang has talked about a doubling of the world’s $1trn data centre infrastructure installed based in the next five years, which will transition from general-purpose to accelerated compute to support AI.

If we are correct, there will also be considerable disruption with many winners from the prior cycle facing a classic incumbents’ dilemma. It is not that they do not understand the new technologies, it is because their business models (like Search in the case of Google) are built around legacy technology and the transition can be painful even for the successful. OpenAI recently unveiled a text-to-video generation tool called ‘Sora’ (the Japanese word for ‘sky’ to reflect its ‘limitless potential’) and showed extraordinary early demos; Adobe Systems (u/w) fell sharply on the day due to the perceived competitive threat. Even among the ‘Magnificent Seven’ year-to-date performance has fractured with NVIDIA and Meta Platforms (Facebook) materially outperforming on the back of AI-related demand and product success while Apple, Alphabet and Tesla are faring less well due to pressures on their core businesses (some due to regulatory headwinds or China exposure). More importantly, the impact of GenAI is uncertain on these businesses, so we have reduced exposure to this group, using them as a source of funds for higher conviction holdings.

While AI capex strength has supported related semiconductor, networking and hardware stocks, investors have been more divided on the implications for software companies. Indeed, semiconductors now make up c10% of the S&P 500 versus software at 12%, the narrowest gap since 2003. The early customer adoption of enterprise application software AI SKUs has been very positive (Microsoft, Adobe Systems and ServiceNow all spoke to their AI SKUs being their fastest-ramping products in history), but the longer-term implications of a new AI software paradigm will also bring challenges to incumbent application software vendors as activities and businesses are rearchitected using AI (we do not know what AI applications will even look like yet).

The fact that Klarna* has already been able to realise significant productivity and cost gains, and the underlying AI models are scaling in a predictable manner, suggests that the future capabilities of AI-powered software will be able to address many more use cases soon. The use of GenAI itself is also creating a much wider ‘attack surface’, providing another route to expose sensitive corporate and personal data, with data and model poisoning representing a new significant threat vector. This could be a major new market: Morgan Stanley estimates that GenAI-assisted cybersecurity could be a >$30bn annual category and only 14% of UBS CIO survey respondents have yet bought security products specifically to address GenAI threats.

Stepping back, the so called “lockout rally” we discussed last month remains in place; it is certainly harder to buy stocks like NVIDIA here if one has not been involved to date. Investors are understandably concerned about the likely volatility of AI-related stocks and the sustainability of AI-related capex, but we are more focused on the existence and continuation of ‘scaling laws’ (predictable changes in model performance due to increasing parameter count/training data). So long as these ‘laws’ continue to hold, model performance will continue to improve, driving a greater range of activities that AI will be able to reach. As always, our aim is to capture the underlying technology revenue and earnings growth potential (in this case infused by AI enablers and beneficiaries), while ameliorating volatility through a diversified portfolio of growth companies.

*not held