Top 3 AI Infra Stocks to BUY NOW in 2026 (Over Nvidia)

Fin Tek•Feb 27, 2026

Summary

Kuran presents a 'bottleneck thesis,' asserting that while Nvidia dominated the initial AI surge, memory is now the critical component limiting further scaling. He highlights that major players like Amazon, Alphabet, Meta, and Microsoft are projected to spend hundreds of billions on AI infrastructure, fueling a 27% compounded growth rate in the HBM market. Kuran emphasizes that traditional memory cycles are being disrupted by the explosive demand for data center HBM, which offers higher margins and tighter supply.

Micron (MU): Kuran identifies Micron as the most direct play on the AI memory bottleneck because their chips are physically required for Nvidia's systems to function. He notes that their newest product, HBM4, launched in February 2026 and is already completely sold out of capacity. Kuran warns that while margins are rising, the stock remains sensitive to any potential slowdown in broader AI spending.

Samsung (SSNLF): Kuran describes Samsung as a scale play within a memory oligopoly that possesses massive manufacturing muscle. He points out that Samsung's stock is relatively undervalued compared to peers, with a cited 42% upside potential according to InvestingPro. He highlights a significant deal for Samsung to supply HBM4 for Nvidia's next-generation 'Vera Rubin' models.

Rambus (RMBS): Kuran views Rambus as a high-potential, higher-risk play that benefits from the increasing complexity of AI server architecture. The company maintains an 80% gross margin by selling intellectual property and interface chips rather than manufacturing physical bricks. Kuran notes that while it has a smaller $10 billion market cap, its high PE ratio of 48 and recent CFO departure are risks to monitor.

Amkor Technology (AMKR): Kuran mentions this company as a bonus play specializing in advanced packaging and testing for AI chips. He acknowledges that packaging is becoming more critical as chips get denser and faster. However, he excludes it from his top three recommendations because the stock currently looks expensive at its current valuation.

Mentioned Stocks

Sentiment: BUYAction: RECOMMENDED

Reasoning: Micron is the most direct way to invest in the AI memory wave as their high-bandwidth memory (HBM) is essential for Nvidia's chips. Their HBM4 product is already sold out of capacity, leading to rising margins and healthy financials.

Loading chart...

AMKR

Sentiment: HOLD

Reasoning: While advanced packaging and testing are critical for the next generation of AI chips, Kuran believes the stock currently looks a little expensive at its current valuation.

Loading chart...

SSNLF

Sentiment: BUYAction: RECOMMENDED

Reasoning: Samsung offers scale and stability within an oligopoly. The stock is currently undervalued with a predicted 42% upside and has secured a major deal to supply HBM4 for Nvidia's future high-performance models.

Loading chart...

RMBS

Sentiment: BUYAction: RECOMMENDED

Reasoning: Rambus is a high-margin IP play (80% gross margins) that manages memory complexity. With a relatively small $10 billion market cap, it has significant upside potential as AI servers require more efficient data movement.

Loading chart...