Ramana Kompella explained that traditional routing methods such as TCP/IP are not suitable for quantum networks because they rely on classical physics. Quantum networks utilize entanglement, where endpoints are connected using entangled photon pairs, allowing for instantaneous information transfer once entanglement is established.
"This is where I think our R&D has done exceptionally well in terms of leveraging existing EUV technology while setting an aggressive technology scaling roadmap," says Kevin Zhang, deputy co-chief operations officer and senior vice president at TSMC. "This is definitely a strength."
"Snowflake gives customers one place to bring their data together, connect the systems they rely on, and turn AI into something that actually helps teams get work done," says Baris Gultekin, VP of AI at Snowflake.
Meta is building these chips because buying AI hardware at scale is expensive, and relying too heavily on external suppliers leaves less room to shape that hardware to its own needs. Building more in-house could help the company keep AI costs in check.
Our custom accelerator business is progressing very well across five customers. Anthropic will soon implement one gigawatt of Broadcom-baked TPUs, and we expect the AI company plans a three-gigawatt deployment in 2027. Meta will install multiple gigawatts of Broadcom's XPU accelerators in 2027 and beyond. OpenAI will deploy over one gigawatt of compute capacity based on custom XPUs in 2027.
