I Found The Next Palantir Setup In AI

Tom Nash•May 1, 2026

Summary

Tom begins by expressing his pride in his investment journey with Palantir, not just for the significant financial returns (a $10,000 investment in January 2023 growing to $210,000 in three years) but for successfully educating retail investors on quality long-term investing. He believes a similar opportunity is emerging within the artificial intelligence (AI) sector, which he identifies as the biggest secular megatrend. Tom emphasizes that successful long-term investing requires ignoring market noise, focusing on quality businesses aligned with global trends, and holding investments regardless of short-term price fluctuations.

He advises against investing in "shiny things" like AI applications, which he compares to gold prospectors who often go broke. Instead, he advocates for the "picks and shovels" approach, investing in companies that facilitate the AI boom, which he categorizes into eight layers: compute, networking, power and cooling, data, control, security, observation, and physical infrastructure. Tom asserts that companies in these foundational layers will "print money over the next 5 years" and that a diversified portfolio of the best companies from each layer holds the potential for returns similar to Palantir. He dismisses skepticism about achieving such returns, pointing to other successful past recommendations like Nvidia (up 1300% since 2020), Arista (up 330% since 2023), Micron (up 61% since March), Vertiv (up 204% since June 2025, likely a typo for 2023 or 2024), and Bloom Energy (up 72% since February). He stresses that "it's never a bad time to invest in a great company" and that playing it safe by holding cash leads to loss of purchasing power.

Tom then details specific company recommendations for each layer:

**Physical Infrastructure Layer (Data Centers, Cloud Providers):** This layer includes companies that own the essential land and roads for AI operations. Tom highlights Google (GOOGL) as the most vertically integrated, possessing a cash cow (Google Search) to finance its cloud expansion (growing at 63%). Amazon (AMZN) is presented as the biggest provider, with AWS showing expanding margins (40%) and revenue growth (30%), offering "the cheapest stock compared to quality." Microsoft (MSFT) is noted for its "best ecosystem" for data centers, with Azure growing at 40%. Oracle (ORCL), though a latecomer, has massive clients and brings new business to the cloud. All are described as "top-notch" and "printing money," making them safe investments.

**Compute Layer (Engine Builders - GPUs, HBM, Semiconductors):** This layer comprises companies building the core processing power. Tom identifies Nvidia (NVDA) as a "quasi-monopoly" and AMD (AMD) as a strong number two in GPUs, both essential. Micron (MU) is highlighted for its dominance in high-bandwidth memory (HBM) with "at least 5 years of total market dominance." ASML (ASML) and TSMC (TSM) are praised as monopolies in building semiconductor manufacturing machines and semiconductors, respectively. Tom notes that these stocks are not expensive, with forward PEs ranging from 5 (Micron) to 29 (AMD, ASML), and are irreplaceable.

**Networking Layer (Road Systems):** This layer focuses on data transmission and communication within data centers. Arista Networks (ANET) is the sole recommendation, lauded for building switches, routers, and software that enable server communication. Tom mentions its impressive financial health (43% operating margin, zero debt, $11B cash, $4.5B free cash flow, 30% revenue growth) and its continuous market share gains in data center networking. He asserts it's "not too late despite the spike" in its stock price.

**Power and Cooling Layer (Fuel System):** This addresses the "biggest choke point in AI right now." Constellation Energy (CEG) is recommended for its nuclear energy solution, offering 24/7, scalable, and cost-effective power. Bloom Energy (BE) provides on-site fuel cells, reducing reliance on the traditional grid. Vertiv (VRT) is crucial for its liquid cooling systems, essential to prevent data center meltdowns. Tom stresses the critical nature of these "picks and shovels" companies in an energy-intensive AI future.

**Control Layer (Central Computer/GPS):** This layer deals with the operating systems that make AI scalable and effective. Tom considers Palantir (PLTR) the "purest control layer competitor," serving as the operating system for AI in large businesses. Despite recent stock dips, he reiterates his strong conviction, emphasizing its long-term potential and past performance (up 1300% since its addition).

**Data Layer (Fuel):** Essential for AI performance, this layer focuses on database management. MongoDB (MDB) and Oracle (ORCL) are singled out as "biggest database monsters" that are "absolutely annihilating it right now" and "critical for the AI mission."

**Observation Layer (Mechanic Software):** This layer involves monitoring and diagnosing AI systems. DataDog (DDOG) is presented as the "monitoring gold standard" with no worthy competitors mentioned.

**Security Layer (Locks and Alarm):** Mission-critical for protecting AI systems and data. CrowdStrike (CRWD) provides endpoint security, acting as a "bouncer on every single computer." Zscaler (ZS) focuses on access security, complementing CrowdStrike. Tom believes these two will be the "best companies to own in cybersecurity for the next 5-10 years."