Investors should prioritize Vast Data on their watchlist as the company prepares for a highly anticipated IPO, offering a high-conviction "pure-play" on AI infrastructure. With storage hardware sold out through 2026, look to major memory manufacturers like Micron (MU), Samsung, and SK Hynix to benefit from the 10x surge in NAND Flash prices. Focus on the "storage layer" of the AI stack, as software-defined storage providers with high margins are becoming as critical to the industry as chipmakers like Nvidia. Vast Data’s ability to store 10 times more data on existing hardware makes it an essential partner for scaling AI operations at firms like xAI and JP Morgan. Diversifying into these infrastructure bottlenecks provides a strategic hedge against the supply chain constraints currently limiting global AI build-outs.

Vast Data (Private)

Vast Data is a high-growth data storage software company that provides the infrastructure necessary for large-scale AI operations. They specialize in managing the massive amounts of data required for AI models to "remember" and process information.

• Market Position: The company is projected to serve one-third of the entire global AI storage build-out this year.
• Core Value Proposition: Vast Data claims its software allows companies to store 10 times more data on the same hardware compared to competitors. This is critical as the cost of NAND Flash (storage chips) has surged up to 10x since August 2025.
• Financial Health:
  • Revenue Growth: Roughly tripling year-over-year.
  • Gross Margins: Approximately 90%, indicating a highly profitable software-centric model rather than a low-margin hardware model.
  • Net Retention: Existing customers are more than doubling their annual spend.
• High-Profile Clientele: Their systems power xAI (Elon Musk’s AI company), JP Morgan, the U.S. Air Force, and the Department of Energy.
• Valuation: Recently raised $1 billion in a Series F round, bringing their valuation to $30 billion.

Takeaways

• Monitor for IPO: The CEO has publicly stated they are preparing for an Initial Public Offering (IPO). Investors should keep this on their watchlist as a "pure-play" way to invest in AI infrastructure beyond just chipmakers like Nvidia.
• Efficiency as a Moat: In a market where storage hardware is sold out through 2026, Vast’s ability to "squeeze" more capacity out of existing hardware makes them an essential partner for AI firms facing supply chain constraints.
• Institutional Validation: The adoption by the Department of Energy and JP Morgan suggests the technology is enterprise-grade and secure, reducing the "startup risk" typically associated with private tech firms.

NAND Flash / Memory Manufacturers

While specific tickers were not mentioned, the transcript highlights a massive supply-demand imbalance in the memory chip sector.

• Supply Constraints: Every major memory manufacturer is reportedly sold out through 2026.
• Price Action: Prices for NAND Flash chips have increased by as much as 10x in a short period due to the AI boom.

Takeaways

• Sector Bullishness: The "sold out" status through 2026 provides high revenue visibility for the memory sector. Investors may want to look into major public memory players (such as Micron, Samsung, or SK Hynix) that benefit from these price hikes.
• Supply Chain Risk: For AI software and hardware integrators, the scarcity of NAND Flash is a significant bottleneck. Companies like Vast Data that mitigate this scarcity through software efficiency are positioned to outperform.

AI Infrastructure & Storage Theme

The discussion shifts the focus from AI "models" (like ChatGPT) to the physical and software "infrastructure" required to run them.

• The "Filing Cabinet" Analogy: AI requires massive "digital filing cabinets" to store prompts, answers, and training data.
• Infrastructure Bottleneck: The primary challenge for AI companies is no longer just processing power (GPUs), but the physical ability to store and retrieve data at scale.

Takeaways

• Diversify Beyond Chips: While GPUs (Nvidia) get the most attention, the "storage layer" is becoming a critical and expensive bottleneck in the AI stack.
• Software-Defined Storage: Look for companies that provide software solutions to hardware problems. Vast Data’s 90% margins suggest that the real value in the storage trade may be in the software that manages the data, rather than the physical disks themselves.