Nvidious position

The market seems to be dominated by the twin beliefs in the invulnerability of the US economy and the massive impact of generative artificial intelligence ( GenAI ). The confidence in the US economy is understandable given that there has been no economic recession for 15 years , barring the special case of Covid in 2020 , and the GenAI excitement fits with the history of potentially transformative technologies , from railways to the internet . This is not the easiest environment for an investment philosophy that looks to back proven and established winners , with earnings that are resilient in tough economic times . When risk is ‘ on ’ and the market is fixated on exponential growth curves , rating stocks on their ‘ AI-ness ’, a portfolio of businesses designed for long-term compounding at reasonable valuations is not in fashion . But what if the prevailing orthodoxy is wrong or starts to unravel ? While the companies we own generally continue to compound and include good potential second-wave beneficiaries of AI , where they control proprietary data and strong market positions , they are not experiencing the rocketing valuations seen by those firstwave AI ‘ winners ’. This narrative is not a surprise to those who invested through the internet bubble of the late 1990s . In recent years , quality has become more conflated with growth , and many quality managers are unabashed in claiming current market winners as quality trophies . The disconnect persists long enough to test the steeliest resolve , to force conversations about whether semiconductors still display cyclical characteristics , and for the conversation to turn to one $ 3 trillion-plus topic : Nvidia ’ s position . Even with its recent pullback , Forbes ’ “ hottest stock of the decade ” ( June 2024 ) – valued at more than 20x sales and above 40x price-to-earnings ( P / E )² – has growth rates more akin to a startup , with earnings expectations having increased a staggering five times versus where they were just two years ago . Nvidia holds a 90 % -95 % market share in data centre graphics processing units ( GPUs ), has earnings before interest and taxes ( EBIT ) margins of 60 %+ and has single-handedly produced the equivalent of 80 % of the market capitalisation rise of the dot-com bubble ³. The current valuation assumes two things : that there will be massive commercial applications of GenAI and that Nvidia ’ s dominance will continue . GenAI does indeed have enormous potential in multiple areas , from coding to customer relations , to image generation and beyond , and we do expect some data-rich constituents of our portfolios to benefit . However , so far at least , most businesses , which are the potential end users , have been experimenting more with GenAI rather than betting heavily as they are often struggling to find clear-cut use cases . There is a stark contrast between the forecasts of close to a trillion dollars of annual GenAI capital expenditures in a few years , and the mere $ 500 million of GenAIrelated revenue that the world ’ s leading IT services company that we hold has reported over the past year . The other threat will come if any of Nvidia ’ s four largest customers ( Microsoft , Amazon , Alphabet and Meta , which currently comprise around 40 % of its revenues ) succeed in their efforts to design a better priced alternative to the H100 or its next-generation follow-up chips Blackwell ( 2025 ) and Rubin ( 2026 ). Economics undergraduate principles spring to mind :