The company’s “persistent exponential progress, the competitive advantages in hardware and software, and the culture and leadership are exactly what we look for”, said Anderson, who last year teamed up with the holding company of Italy’s Agnelli family to launch Lingotto Investment Management, where he runs a US$650b fund whose largest position is the US chipmaker.
Anderson is best known for the almost four decades he spent at Baillie Gifford. There he ran its flagship Scottish Mortgage Investment Trust, which first bought Nvidia in 2016, and helped turn the Edinburgh-based private partnership into an unlikely star of tech investing.
When Scottish Mortgage initiated a position in Nvidia, “it wasn’t clear what would be the main driver – we didn’t decide if it would be gaming, crypto, autonomous driving or AI but left it to the course of events”, said Anderson.
He added that the big difference between the semiconductor manufacturer and some of his other successful bets is that “Amazon, Tesla etc didn’t start from highly profitable and dominant positions but had to get there”.
A crucial influence on Anderson and Baillie Gifford’s investment process has been academic Hendrik Bessembinder, who found that over many decades just 4% of stocks accounted for all the net wealth creation – providing the basis for their belief that fund managers should seek to identify companies that are extreme winners.
Anderson outlined why Nvidia falls into this category in a letter to investors this year.
He wrote that real growth in data centre AI chip demand appeared to be running at about 60% a year. Looking ahead over the next decade, he said 10 years of 60% growth in data centre revenue alone and with unchanged margins would translate into earnings of US$1350 per share and free cashflow of about US$1000 a share. Assuming a 5% free cashflow yield, an Nvidia share might be worth US$20,000 in 10 years, which translates to a market capitalisation of US$49t. Anderson put the probability of this type of outcome at 10-15%.
The current combined market capitalisation of all the companies in the S&P 500 is roughly US$47t.
“It is the long duration of the development of GPU usage in AI – and not just AI – from excitement, through potential pauses, to transformation of industries that is most important to us,” said Anderson.
He added that the path was likely to be volatile and that he would not be surprised if Nvidia had one or more drawdowns of 35-40% – “that’s what happens and I hope we’d buy more in that event”.
Nvidia currently trades at more than 47 times its estimated earnings per share for the coming year and is responsible for almost 30% of the S&P 500′s 17.7% gain this year.
The growing sway of Nvidia and the largest tech “megacaps” over broader stock market indices has provided challenges for fund managers who do not hold them. For example, Terry Smith’s global fund lagged behind its benchmark in the first half of the year after choosing to avoid the chipmaker because “we have yet to convince ourselves that its outlook is as predictable as we seek”.
Addressing the question of whether generative AI has been overhyped, Anderson said “the narrow generative AI for basic and consumer tasks may well be overblown, but we see the big issue as whether it can solve serious problems in 10 years”, including autonomous driving, robotics and drug discovery. “And in that sense it’s the opposite of hype... Nvidia is quietly but firmly leading in backing and providing these areas.”
Written by: Harriet Agnew and George Steer in London
© Financial Times
