Nvidia and Palantir launched a sovereign AI engine for U.S. agencies combining Nemotron models with Foundry, AIP and Apollo for enforced data isolation and auditability. ByteDance’s plan to mass-produce CPUs by H2 2027 poses a headwind to Nvidia’s China accelerator sales, even as Goldman Sachs expects another AI-driven earnings surge.
Nvidia and Palantir integrate Nemotron open models with Palantir’s AIP, Foundry, Ontology and Apollo to build an AI engine tailored for U.S. government environments. The solution enforces explicit data authorization, customer-specific isolation, secure perimeters, data portability and full auditability while enabling continuous model improvement with new data and feedback.
ByteDance aims to finalize its next-gen in-house CPU design by early 2027 and start mass production in H2 2027 to support AI workloads like its Doubao chatbot and Seedance model. This strategy could reduce reliance on Nvidia GPUs in China, layering on existing export-control headwinds for advanced accelerators.
Goldman Sachs forecasts that Nvidia’s earnings growth will be driven by ongoing AI demand in the coming quarters, highlighting robust GPU adoption across cloud and enterprise segments. The bank expects AI-related sales to sustain double-digit revenue expansion without specifying absolute figures.
