AIToday
AI Business & IndustryDIGITIMES AsiaPublished: Sep 1, 2026, 16:01 JST1 min read

AI optical interconnects move into racks as 400G/lane and CPO mature

AI optical interconnects move into racks as 400G/lane and CPO mature

Key takeaway

  • AI scaling is moving interconnect needs from chips to racks and beyond.

  • Optical interconnects at rack level are now key.

  • 400G/lane and CPO maturity are the main factors to watch.

3 Key Points

  1. What happened

    AI system scaling has pushed interconnect requirements inside data centers from chips and boards up to racks, clusters, and even inter-data-center links, according to DIGITIMES.

  2. Why it matters

    This shift means optical interconnect technology is becoming critical at the rack level, with 400G/lane speeds and co-packaged optics (CPO) emerging as key variables for AI infrastructure.

  3. What to watch

    How quickly 400G/lane and CPO mature will likely determine the pace at which rack-level optical interconnects are adopted in AI data centers.

Ask the AI about this article →

Context & Analysis

DIGITIMES observes that as AI systems scale, the demand for interconnectivity inside data centers has expanded beyond individual components. This progression from chips to racks and beyond signals a growing reliance on optical solutions at higher levels of the infrastructure stack. The mention of 400G/lane and CPO as key variables suggests that the industry is at a decision point where the maturity of these technologies will shape the next phase of AI data center buildouts. For business readers, this implies that investments and architectural choices in AI infrastructure will increasingly depend on optical interconnect performance, potentially affecting supply chains and technology roadmaps.

FAQ

What are the key variables for rack-level optical interconnects?
The maturity of 400G/lane technology and co-packaged optics (CPO) are the key variables.
How far have interconnect requirements expanded?
They have expanded from chips, packages, and boards to racks, clusters, and eventually inter-data-center connections.
DIGITIMES AsiaRead Original Article

Get the latest AI Business & Industry news every morning

For example, today's edition would include:

  • Z.ai runs GLM on 100,000 Chinese AI chipsDIGITIMES Asia · 1h ago
  • Nvidia revives Rubin CPX chip with major redesignYahoo Finance AI · 1h ago
  • Palantir stock near fair value; AI data debate weighsYahoo Finance AI · 1h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleNvidia revives Rubin CPX chip with major redesign