Cerebras expands globally, targets NVDA & MSFT

Cerebras' expansion into Europe and Canada is a strategic move to bring AI compute closer to customers, addressing the rising demand for high-speed inference. The company's infrastructure, which does not require gigawatt-scale sites, allows flexibility in choosing locations, unlike traditional training clusters. This approach broadens its addressable market and positions it to compete with larger players like NVIDIA and Microsoft.

The revenue contributions from international customers, particularly the Mohamed bin Zayed University of AI, underscore the importance of sovereign AI initiatives. These customers require more control over data and infrastructure, which Cerebras supports through on-premises, cloud, and partner-cloud offerings. The AWS collaboration, expected to go live in Q1 2027, could further expand its reach, although the $25 billion in remaining performance obligations do not yet include this partnership.

The company's manufacturing capacity increase of more than 10-fold in 2026 is a key enabler for its geographic growth, supporting the infrastructure needed for additional deals. With discussions with other hyperscalers potentially generating revenues in mid-2027 and ramping through 2028, Cerebras is positioning itself for sustained international expansion, though the full impact of these efforts remains to be seen.