You’re reading an issue of “The AI Economy,” my newsletter exploring the forces shaping the AI era—tracking how AI is rewriting business, work, technology, and culture. Subscribe to get expert insights and curated updates delivered straight to your inbox.
The Allen Institute for AI (Ai2) said Thursday that its federally backed AI computing cluster is now online, marking the first milestone of a $152 million program to build open AI models for scientific research. The project, called the Open Multimodal AI Infrastructure for Science (NSF OMAI), is backed by the U.S. National Science Foundation (NSF) and NVIDIA, whose Blackwell Ultra GPUs power the new infrastructure. Ai2 said the cluster will be used to develop a fully open AI ecosystem aimed at accelerating discovery in fields like materials science, biology, and energy.
Last August, Ai2 was awarded funding to lead OMAI through an NSF Mid-Scale Research Infrastructure grant. The project aligns with the White House AI Action Plan, which prioritizes accelerating AI-enabled science and U.S. leadership in open AI development. And while Ai2 is associated with OMAI, it’s not alone: The lab is working alongside other principal investigators from the University of Hawaii at Hilo, the University of New Hampshire, the University of New Mexico, and the University of Washington.
“At a time when access to advanced AI systems is increasingly concentrated among a small number of companies, bringing this hardware infrastructure online represents a critical step for us,” Noah A. Smith, principal investigator at NSF OMAI and senior research director at Ai2, said in a blog post. “Our goal is to accelerate a truly open technology ecosystem with broad impact, developing fully-open AI systems, resources, and tools that strengthen AI research and support continued U.S. leadership in the field.”
NSF OMAI reflects Ai2’s long-standing commitment to fully open AI development. In closed systems, compute spent on experiments and iteration typically yields a single commercial product. However, when that same work is made open, Ai2 argues it continues to generate value “long after training ends.” The new cluster is built around that philosophy, prioritizing how effectively capacity is used and shared rather than the sheer size of its computing footprint. It’s also deployed and managed by Cirrascale Cloud Services, the infrastructure platform provider that Ai2 already partners with to distribute its Olmo, Molmo, and Tulu AI models.
“By investing in open, shared resources, we are enabling scientists and researchers across disciplines to build, test, reproduce, validate, and advance AI systems,” Wendy Nilsen, the deputy directorate head for NSF’s computer and information science and engineering directorate, said. “This work accelerates discovery, strengthens scientific rigor through replicability and transparency, and reinforces U.S. leadership in the field.”
Ai2 said NSF OMAI is already producing results. For example, Molmo 2 introduced video understanding, pointing, and object tracking to its multimodal model family—with an 8B-parameter model outperforming the original 72B Molmo on key benchmarks. A follow-up release, MolmoPoint, replaced text-coordinate outputs with a token-based grounding mechanism, achieving state-of-the-art accuracy on spatial reasoning tasks. And on the language side, Olmo Hybrid combined transformer attention with linear RNN layers, matching prior models while using roughly half the training data.
Taken together, they represent the “breadth of research” that NSF OMAI is capable of accelerating across A2’s language modeling programs, creating not only models, but also “open artifacts that other teams can inspect, adapt, and build on.”
