Susquehanna's Mehdi Hosseini: AI spending is ‘spend now, ask ROI later'

CNBC TelevisionAbout 4 min readSep 24, 2025Watch original
THE SUMMARYAI-generated

Key Concepts:

  • High Value Added (HVA) Memory Products
  • AI Server Memory Demand
  • Customer Concentration in AI Supply Chain
  • DRAM (Dynamic Random-Access Memory)
  • Margin Expansion
  • Semiconductor Capital Equipment (Semi Cap Equipment)
  • Vendor Financing
  • ROI (Return on Investment)

1. Micron's Positive Outlook and Pricing Power:

  • Susquehanna has a "buy" rating on Micron (MU) with a price target of $200.
  • The positive outlook is driven by a better pricing environment due to a richer mix of high value-added products and disciplined spending.
  • Unlike previous cycles driven by distributors and OEMs, the current cycle is driven by Nvidia, AMD, and Broadcom/Avago.
  • Nvidia, which accounts for about 80% of the AI server cost, is a major buyer of memory from Micron.
  • The future of AI compute requires advanced memory, giving Micron pricing power.
  • This cycle is expected to be sustainable throughout 2026 and potentially into 2027.

2. Customer Concentration and AI Supply Chain:

  • There is customer concentration throughout the AI supply chain, with Nvidia being a key player.
  • Micron shouldn't be penalized for this concentration, as it reflects the downside risk to overall AI spending.

3. Micron's Potential for Margin Expansion and Earnings:

  • The next generation of AI chips (Rubin) will require even more advanced DRAM.
  • Micron has the potential to expand margins towards 55% or higher.
  • The earning power of Micron is estimated to be around $20 per share, supporting a $200 stock price even without multiple expansion.

4. Semiconductor Capital Equipment (Semi Cap Equipment) Preferences:

  • Susquehanna has a "hold" rating on Applied Materials (AMAT) due to challenges in China, where local equipment suppliers are catching up.
  • Lam Research (LRCX) is favored because it is overindexed to memory spend, which is expected to pick up in the second half of 2026 due to supply constraints.
  • Lam Research is also benefiting from increased market share and a new material used in making semiconductor chips.
  • ASML is another favored name, as Taiwan Semiconductor Manufacturing Company (TSMC) is expected to be its largest vendor in 2026, and ASML is overindexed to Taiwan.

5. Vendor Financing Concerns:

  • Vendor financing, where vendors invest in projects, is a concerning trend reminiscent of the solar industry in 2008-2009.
  • While AI compute is more sophisticated, vendor financing could backfire, leading to consolidation and a pause.
  • The expected pause, initially anticipated in 2026, is now pushed out to 2027.
  • There is a strong trend of "spend, spend, spend," with the ROI question being addressed later, potentially in 2027.

6. Notable Quotes:

  • "Unlike prior cycles, it is Nvidia, it is AMD, it is Broadcom Avago that are actually doing the buying."
  • "The future of AI compute requires advanced memory. And that's where the premium comes in."
  • "There is a very strong trend in spend, spend, spend. And then you ask the ROI question later."

7. Logical Connections:

  • The demand for advanced memory in AI servers drives Micron's pricing power and positive outlook.
  • Customer concentration in the AI supply chain is a risk factor but also reflects the overall demand for AI.
  • Margin expansion and earnings potential support the $200 price target for Micron.
  • Preferences in semi-cap equipment are based on exposure to memory spend and market share gains.
  • Vendor financing is a concern that could lead to a pause in spending and consolidation in the future.

8. Synthesis/Conclusion:

The analyst maintains a positive outlook on Micron due to the strong demand for advanced memory in AI servers, driven primarily by Nvidia. This demand gives Micron pricing power and the potential for margin expansion. While customer concentration and vendor financing are risks, the overall trend of spending on AI compute is expected to continue, pushing out any potential pause to 2027. The analyst favors Lam Research and ASML in the semi-cap equipment space due to their exposure to memory spend and market share gains.

AI summaries can miss context or contain errors. Check important details against the original video.

MAKE IT YOURS

Read. Remember. Reuse.

Free tools

Go a little deeper.

Have a question about this video? Load its transcript to open the video chat.