Valiant Capital Management has aggressively ramped up its bet on artificial intelligence.
The hedge fund firm headed by Tiger Grandcub Chris Hansen said that in the first quarter it made two significant new AI investments, one of which is now the firm’s second-largest long position, according to the first-quarter client letter, obtained by Institutional Investor.
Valiant also significantly boosted the shares of AI superstar Nvidia, previously a smallish position in the hedge fund’s portfolio. In the first quarter, Valiant boosted its stake by more than 90 percent, making it the firm’s sixth-largest U.S.-listed common stock long, according to the 13F filing made public Wednesday.
“Despite this appearing to be a somewhat consensus view, we think that most investors are vastly underestimating the scale of investment required,” Hansen stressed in the letter. “We believe AI-enhanced applications are already generating substantial productivity gains in existing business workflows — with the clearest current examples being coding and data science. And the scale of productivity enhancement and cost savings [is] so substantial that the pace of adoption is likely to explode in the coming quarters and years.”
After gaining 1 percent last month, Valiant is up 6.1 percent for the year through April, according to a separate monthly email to clients that was seen by II. In the first quarter, longs kicked in 11 percent to gross gains, shorts detracted 3.86 percent, and macro bets cut into performance by 1.3 percent, per the letter.
In the first-quarter report, Hansen reminded clients that as in previous innovation cycles, new tech business models are emerging daily. “From OpenAI’s video generator Sora to Perplexity AI’s reimagined search engine to Figure’s new AI robots to the multitude of applications exploding in content creation, the number of native, AI-first companies coming to market is unprecedented in scale and implication relative to anything we have seen since the dotcom boom of the late 1990s and early 2000s,” he asserted.
For AI to realize its potential, Hansen said, infrastructure for both model training and inferencing must be ramped up.
“And researching AI infrastructure on these two vectors (model training and inferencing) has led us to a more confident view of the scale of the AI market opportunity, the real business value that can be generated, and the necessity of larger, nonconsensus infrastructure investments needed to realize that value,” he added.
So in the first quarter, Valiant added Taiwan Semiconductor Manufacturing Co. and Broadcom to its portfolio. It also increased its existing but small Nvidia position and added “starter positions” in a few other names it said it is digging into “to develop a clearer view as to whether they deserve a more significant spot in the book.”
At the end of the first quarter, TSMC was Valiant’s second-largest long position, behind Eli Lilly and ahead of Microsoft, Salesforce, and India-based HDFC Bank, the letter says. Hansen also noted that Valiant used the market weakness in April to add to its exposure to these companies.
The hedge fund manager explained that the firm has long admired TSMC’s long-term business strategy, widening competitive position, and exceptional execution. “We believe TSMC is one of the best businesses in the world and is the only provider of truly leading-edge semiconductors,” Hansen said. “As such, we are confident the substantial performance, power, and cost efficiencies of their leading-edge processors should mean the global growth of AI infrastructure will overwhelmingly benefit them.”
As for Broadcom, Hansen said that over the last quarter, Valiant became enthusiastic about the company’s opportunity to capitalize on future demand surge for inferencing technology.
“Along with our increased bullishness on the size of the AI market, we see Broadcom especially exposed to two areas of substantial future growth: their leadership in custom ASIC designs and their long-standing networking leadership,” he said.
Hansen asserted that custom ASICs — processing units designed for a specific use case — will obtain “a meaningful market share of AI compute over time.”
He explained: “As the type and predictability of AI use cases and workloads become clearer, we believe the importance of cost and power to lowering the overall cost per inference query will lead to a large increase in custom designs. As Broadcom is the clear leader in custom ASICs with the biggest deployments (including a decade developing AI-related designs for Google), we believe that as other very large customers look to deploy custom ASICs for inferencing, they too will look to Broadcom.”
Hansen also asserted that Broadcom’s networking leadership “will shine through” at an increasing rate as a broader array of hardware/compute is deployed for AI inferencing and that as the clear market leader in networking chips, Broadcom stands to “disproportionately benefit from this trend over the longer term.”
He reminded clients that TSMC and Broadcom are the types of businesses that have long been the core of Valiant’s investment strategy: extraordinary companies with high returns on capital, large and expanding moats, and solutions for many of their customers’ biggest needs.
“While these are some of our newest investments that are poised to benefit from AI, they will not be our last,” Hansen insisted. “As we believe this could be a megatrend that will play out for decades to come, our research is continuing in earnest.”