TickerTain
$SPCX····$NVDA····$MSFT····$GOOGL····$ANTHROPIC·$TSLA····$CRWV····$NBIS····$HIMS····$TSM····$ASML····$FIGURE·$LYFT····$UBER····$SNOW····$META····$ELEVENLABS·$CURSOR·$COGNITION·$OPENAI·$AAPL····$RKLB····$XAI·$AMZN····$AMD····$000660.KS····$INTC····
← newsroom
$AMD · Why next-gen AI scale-up needs CPO
now playing · $AMD
$AMD···neutral· lowunknown
AMD Infinity Fabric on-package interconnects precedent for CPO integration
AMD's Infinity Fabric on-package interconnects (used since Zen 1 through Zen 5 to connect compute and I/O dies) demonstrate the feasibility of on-package high-speed electrical int…
$TSM···neutral· lowunknown
TSMC SOIC-MH advanced packaging enables 2.5D stacking for co-packaged optics
TSMC's latest SOIC-MH packaging technology enables 2.5D stacking with hybrid bonding, allowing optical engines to be placed close enough to ASICs to eliminate SERDES and DSPs, rep…
$AVGO···neutral· mediumunknown
Broadcom CPO solutions create vendor lock-in risk for hyperscalers
Hyperscalers face vendor lock-in with CPO because the optical engine is co-packaged with the switch ASIC — buying Nvidia hardware means buying Nvidia's CPO solution, buying Broadc…
$NVDA···bullish· highunknown
Nvidia roadmap pushes copper to limit then adopts CPO for massive scale-up clusters
Nvidia's Blackwell NVLink 72 rack achieves 72-GPU scale-up via copper; next-gen Rubin Ultra NVL 576 (8 racks, 576 GPUs) and Kyber NVL 1,152 will combine copper intra-rack with CPO…