1. Navitas secured a major AI data center partnership with Nvidia to supply next-gen GaN and SiC power semiconductors for 800V HVDC systems; 2. The collaboration positions Navitas as a leader in enabling more efficient, cooler, and cost-effective AI infrastructure by reducing copper reliance; 3. Despite a 150% pre-market surge, the author argues this marks the beginning of long-term growth as high-voltage solutions gain industry traction.
Recent #AI Data Centers news in the semiconductor industry
➀ The adoption of OCS optical switching in AI data centers can significantly reduce power consumption, costs, and latency, but the market is still in its early stages with great uncertainty in its development trend.
➁ Google's OCS structure for AI is simpler compared to the unblockable WSS optical switching in trunk ROADM, but it is a blocking structure that requires overall port traffic switching.
➂ The choice of OCS type for AI reconstruction depends on the requirements for reconstruction speed and scale. Larger mechanical structures can be used for large-scale exchange without high-speed requirements, while smaller arrays of MEMS, liquid crystal, or piezoelectric technologies can be chosen for lower-scale applications.
1. Nvidia reported strong Q2 earnings but experienced a growth slowdown, leading to a 5% drop in after-hours trading. 2. The data center segment drove significant revenue growth, but Nvidia's heavy reliance on this unit poses risks. 3. Gross margins declined sequentially, and the new $50 billion buyback authorization is unlikely to significantly impact Nvidia's share count.