Summary
Highlights
Tier three addresses the crucial need for power and cooling in AI data centers. The massive heat generated by AI chips and their electricity consumption are major challenges. Data centers currently use 4% of US electricity production, projected to triple to 12% in five years. This problem creates profitable opportunities for companies like Eaton Corporation (45% increase in data center sales with an Nvidia partnership), G Venova (600 million in data center orders), Siemens Energy (power plants), and ABB (electrical equipment). The speaker emphasizes that future AI chips will demand even more power, making these companies vital.
Felix outlines an action plan: 1) Audit your current portfolio for AI exposure. 2) Decide your allocation to AI and within each tier. 3) Learn when to sell using a structured exit strategy (offered via a free training session). 4) Build positions gradually, avoiding all-in investments. 5) Regularly review your portfolio, quarterly for investors and weekly for traders. 6) Continuously learn and improve investment skills, understanding key criteria for good companies, money flow, and profit-taking timing. He stresses the importance of not letting significant profits slip away, offering a promise to fulfill this in his training session.
The video opens by addressing investors in Nvidia, tech ETFs, and AI-focused funds, revealing that a significant portion of AI data center capital expenditure (estimated at $7 trillion by the decade's end) is flowing into companies beyond just chip manufacturers. Only 20% goes to chip companies, while the other 80% is directed towards less-known companies involved in silicon manufacturing, server building, and powering AI data centers that consume immense electricity. The speaker, Felix Pri, a former investment banker and founder of The Goat Academy and Tradevision.io, aims to educate regular investors on protecting wealth and identifying overlooked opportunities. He emphasizes researching the semiconductor supply chain and institutional findings.
Felix introduces three critical areas: why the "pick and shovel" strategy consistently outperforms direct investments in tech cycles (from the gold rush to cloud computing), the three-tier AI infrastructure stack (chip manufacturers, system integrators, power brokers), and specific companies poised to benefit. He illustrates the "pick and shovel" strategy with examples like Cisco during the dot-com bubble and infrastructure companies in cloud computing, noting their superior returns with less risk compared to the highly speculative application companies. He warns about the current high valuation risk and large multiples of big tech stocks, making up 33% of the S&P, and suggests that true growth lies in the ecosystem powering the AI revolution.
Tier one focuses on the companies that actually manufacture the chips that designers like Nvidia create. Key players highlighted include TSMC, the world's most advanced chip foundry with 34% year-on-year revenue growth, and ASML, which holds a monopoly on lithography machines essential for advanced chip production. Other equipment manufacturers like Applied Materials and KLA Corporation, crucial for quality control in precise chip manufacturing, are also mentioned. The speaker notes the lower valuation multiples of these manufacturers compared to chip designers, suggesting they are worth investigating.
Tier two comprises companies that integrate chips into functional AI servers. Super Micro is highlighted for building AI servers, specializing in liquid cooling, and holding a 23% market share. Broadcom is identified for building custom AI chips and networking solutions, with a staggering 220% revenue growth, supplying giants like Google and Meta. Marvell Technology, with 78% growth in custom silicon and electro-optics for data centers, is also mentioned. The speaker points out that these companies make money regardless of whether Nvidia or AMD wins, working with all major hyperscalers, and have significantly lower valuations than Nvidia.