AUTO-UPDATED

Everyone Wants to Build AI Apps — I’m Betting on What Powers Them. Here’s Why.

The artificial intelligence industry is shifting its focus from model training to inference, driving massive capital investment into specialized chipmakers that prioritize speed, efficiency, and production-scale performance.

Key Points

  • Cerebras Systems achieved a $95 billion market cap following a successful IPO that raised $5.55 billion to advance wafer-scale compute technology.
  • Groq secured $650 million in funding to expand its inference-focused AI cloud business, signaling a strategic pivot away from general-purpose training hardware.
  • Positron raised $875 million in a September 2026 funding round to develop specialized chiplet architecture designed to compete with Nvidia.
  • Tenstorrent reached a $5.76 billion valuation after closing $1.4 billion in Series E funding to scale its specialized silicon infrastructure.
  • d-Matrix maintains a $2 billion valuation, focusing on digital in-memory compute solutions specifically optimized for high-demand inference workloads.

Why it Matters

The transition toward inference-focused hardware suggests that long-term value in the AI sector is moving from the volatile application layer to the underlying infrastructure. Investors and founders are increasingly prioritizing specialized silicon providers that offer the efficiency required to run AI models at scale, rather than relying on general-purpose hardware.
Entrepreneur Published by Giuseppe Cicatelli
Read original