Chip stocks will get stronger as the year progresses, says top tech investor Dan Niles

Watch on YouTube ↗  |  April 07, 2026 at 21:44  |  4:25  |  CNBC

Summary

  • The AI cycle has shifted from the initial training phase to inference, and now into a new "age of AI" or "agentic" phase, which requires significantly more compute/tokens.
  • Token production growth has accelerated from ~20% before OpenAI's late-January event to over 130% afterward, driving a massive increase in compute demand.
  • Semiconductor ("chip") stocks are in a temporary lull but are expected to get "a lot stronger" throughout the rest of the year due to this agentic AI shift. Investors will need to be selective within the sector.
  • NVIDIA is seen as similar to Cisco in 1997-98, experiencing significant intra-year pullbacks but finishing the year much higher. It is expected to end the current year higher than today's level.
  • A positive sign for NVIDIA is its stock closing up on a day when a major competitor (Broadcom) announced a big deal, showing resilience.
  • Market participants are becoming more selective around eventual AI winners, creating dispersion.
  • For Meta (META), a capex cut focused on non-core "science projects" (e.g., Reality Labs) would likely be viewed positively by the market, as seen in its last earnings reaction.
  • In contrast, a capex cut from a major cloud provider (AWS, Google, Azure) would be viewed negatively, as high multiples are predicated on them all being winners.
  • The AI model market is showing signs of pricing pressure and differentiation: Anthropic is raising prices while OpenAI is cutting them, suggesting a more competitive and selective environment.
Trade Ideas
Dan Niles Founder & Portfolio Manager, Niles Investment Management 2:36
The speaker explicitly states "I think NVIDIA ends the year higher than where it is today." He cites its positive price action on a day with potentially negative competitor news as an encouraging near-term sign and draws a direct historical analogy to Cisco's performance in 1997-98. The transition to "agentic" AI is driving a massive increase in token production and compute requirements, which sustains and grows demand for NVIDIA's hardware. LONG due to the expectation of a higher stock price by year-end, supported by fundamental demand drivers and historical precedent for digestion phases preceding strong finishes. OpenAI may have "a lot more problems than investors currently think." A broader market reassessment and increased selectivity around AI winners could also pose a headwind.
Up Next

This CNBC video, published April 07, 2026, features Dan Niles discussing NVDA. 1 trade idea extracted by AI with direction and confidence scoring.

Speakers: Dan Niles  · Tickers: NVDA