High US cloud service suppliers proceed to steer development within the AI server market, in line with TrendForce
In sum – what to know:
AI momentum – International AI server shipments are projected to rise 24.3% in 2025, barely beneath forecasts because of U.S. export restrictions and geopolitics.
Cloud methods – AWS, Google, Microsoft, Meta and Oracle are increasing AI infra with various mixes of Nvidia GPUs and in-house chips.
OEM shifts – Server distributors are revising H2 2025 methods, whereas sovereign cloud and regional AI initiatives are boosting demand in EMEA, Asia.
International demand for AI servers stays sturdy in 2025, with ‘tier-one’ North American cloud service suppliers (CSPs) persevering with to steer market development, in line with new evaluation from market analysis outfit TrendForce. Risng demand can be being supported by tier-two information facilities and sovereign cloud deployments throughout Europe and the Center East, it stated.
Regardless of sturdy momentum for AI servers at a world stage, TrendForce has lowered its annual forecast marginally because of rising geopolitical tensions and ongoing U.S. export controls, mainly affecting the Chinese language market. As a consequence of this, the worldwide AI server shipments at the moment are projected to develop 24.3% year-over-year — nonetheless nicely into double-digit territory.
Among the many large U.S.-based hyperscalers, Microsoft continues to prioritize AI infrastructure investments, favoring Nvidia GPU-powered techniques. The corporate’s internally developed ASICs — particularly the next-generation Maia chips — are progressing extra slowly, with quantity ‘ramp’ anticipated in 2026, in line with TrendForce. Microsoft’s AI push has additionally slowed its purchases of general-purpose servers, it added.
In the meantime, Meta is increasing each its AI server capability and general-purpose server footprint. New information middle completions have pushed sturdy AMD-based server orders. Meta’s in-house AI chip program can be rising, with MTIA chip shipments anticipated to double by 2026, in line with the report.
TrendForce highlighted that Google’s server development has been primarily pushed by sovereign cloud initiatives and recent information middle capability in Southeast Asia. “Google has begun mass deployment of its AI inference-focused TPU v6e chips, which have turn into mainstream within the first half of 2025,” it stated.
Amazon Internet Companies (AWS) is targeted on scaling its in-house Trainium v2 platform whereas creating a number of variants of Trainium v3, scheduled for mass manufacturing in 2026. AWS is anticipated to steer all U.S. CSPs in in-house AI chip shipments this yr, doubling its 2024 volumes, the analysis agency discovered.
“Oracle, in comparison with the opposite 4 CSPs, is extra centered on buying AI servers and in-memory database (IMDB) servers. In 2025, Oracle plans to step up AI server infrastructure deployment and combine its core cloud database providers with AI functions. In response to sovereign cloud initiatives within the US, the corporate has additionally seen elevated demand for NVIDIA’s GB Rack NVL72 options,” stated TrendForce.
“Moreover, many server enterprise OEMs are reassessing their market methods for the second half of 2025 because of current adjustments in worldwide tariff insurance policies. At present, TrendForce estimates that complete server shipments—together with each general-purpose and AI servers—will see a year-on-year development of roughly 5%, in keeping with earlier projections,” it added.
