Close-up of a computer circuit board. Photo: Ivan Chumak / Pexels
2 min read

Positron banks $875M to beat GPUs at memory bandwidth

Positron AI just raised $875 million to prove inference doesn't need to choke on HBM forever. THE GIST Positron closed an $875M Series C at roughly a $5B valuation for memory-first Asimov inference chips, per SDxCentral. The LPDDR5X design claims over 90% memory-bandwidth use versus under 30% for GPUs, with TSMC 3nm tapeout targeted by end-2026. THE TAKE This is a bet that inference economics will be decided by memory walls, not FLOPS bragging rights. HBM is scarce, pricey,

THE HOOK

Positron AI just raised $875 million to prove inference doesn't need to choke on HBM forever.

THE GIST

Positron closed an $875M Series C at roughly a $5B valuation for memory-first Asimov inference chips, per SDxCentral. The LPDDR5X design claims over 90% memory-bandwidth use versus under 30% for GPUs, with TSMC 3nm tapeout targeted by end-2026.

THE TAKE

This is a bet that inference economics will be decided by memory walls, not FLOPS bragging rights. HBM is scarce, pricey, and power-hungry; shipping LPDDR5X into Titan servers is the adult move. If Asimov actually ships in 2H 2027, Nvidia's inference tax gets a real competitor. If tapeout slips, it's just another beautiful slide deck.

OPINION

The cluster story—up to 16,384 chips and ~128 Tb/s chip-to-chip—is the part to watch, not the founder mythology. Sovereign money from Qatar and Cisco in the cap table says this is treated as infrastructure. Full breakdown in the SDxCentral report.

  • Story type: Breaking news
  • Impact: High
  • Confidence: High (official announcement)
  • Audience: Business
  • What changed: Positron raised $875M at ~$5B for memory-first Asimov inference silicon
  • What to watch: TSMC 3nm tapeout timing at end of 2026 and Titan production ramp in 2H 2027
  • Reading time: 2 minutes

Continue reading