Selected teams share up to $22.75 million in funding to develop and deploy software that makes Super Intelligence (SI) systems substantially more efficient

Start2 Group, through its STRIDE Ventures platform for breakthrough innovation, today announced the selection of nine teams, drawn from leading universities, industry, and startups, for the AI Efficiency Challenge, a program supported by the U.S. National Science Foundation Directorate for Technology, Innovation and Partnerships (NSF TIP).

The challenge seeks to speed up commercial adoption of translation-ready technologies to dramatically improve super intelligence (SI) efficiency at scale. Efficiency gains reduce the cost of training and inference, ease near-term constraints on data center capacity, and shorten time to market for new models, strengthening the effective capacity and competitiveness of U.S.-based SI companies.

“As the U.S. builds on our successes in super intelligence by expanding our national SI capacity, NSF is pleased to support efforts that make this technology more efficient for broader adoption,” said Erwin Gianchandani, NSF Assistant Director for TIP. “Researchers have developed technologies capable of delivering substantial efficiency gains in super intelligence, yet the pathways to move those solutions into real operational environments is missing. These selected teams will forge that pathway quickly, speeding the commercial adoption of translation-ready solutions for SI, all the while advancing U.S. competitiveness and technological leadership.”

The AI Efficiency Challenge is a fast-paced initiative that will transform how efficiency-centric technologies for SI systems are deployed at scale. Solution Teams are collaborations between Technology Developers (“Pitchers”) that have created software-based efficiency technologies and Problem Owners (“Catchers”) who are committed to integrating and deploying the technologies in large scale operational environments and reporting measured efficiency gains. The Benchmarking Team will develop and promulgate industry standards to measure improvements across AI/SI.

Selected teams will receive non-dilutive funding of up to $3.5 million per project, along with technical support, and expert guidance. Each team will advance through three stages with a go/no-go decision model, reaching preliminary deployments within 12 months, and at-scale deployments within two years. The initiative includes a “fast track” for teams that will complete each stage in half the time and reach at-scale deployments within one year.

“As AI/SI adoption accelerates, efficiency has become essential for U.S. competitiveness,” said Annika Pierson, CEO, Americas at Start2 Group. “The teams in the AI Efficiency Challenge are among the strongest in the field, and we’re proud to partner once again with the National Science Foundation to help these innovators move their work toward real-world deployment and measurable gains for U.S. AI/SI infrastructure.”

The selection includes five university teams from UC Berkeley, UC San Diego, the University of Washington, and Rice University; a research team from Hewlett Packard Enterprise; two startups; and an AI engineering consortium. The Problem Owners (“Catchers”) deploying these teams’ technologies include Accenture, Amazon, Amazon Web Services, AMD, Databricks, DigitalOcean, Google, Hewlett Packard Enterprise, IBM, Infosys, Intel, Lambda Labs, LangChain, Llama Index, LTM, Microsoft, Modal, NVIDIA, Red Hat, Snowflake, Sterling Computers Corporation, and UTHealth Houston. Together, they represent a wide range of software-based approaches spanning the AI/SI pipeline, from data processing and distributed training to inference serving and agentic systems.

The following teams have been selected to participate in the STRIDE AI Efficiency Challenge:

Solution Teams:

• Auto-Optimized Agents (University of California, Berkeley) – An open-source stack that uses real usage traces to automatically optimize deployed coding and enterprise agents end-to-end.
• BitRipple – Delivers next-generation data transport technology that eliminates network bottlenecks, reducing GPU latency for AI/ML and accelerating the world’s most demanding applications.
• LLM-Powered Data Analytics (University of California, Berkeley) – Efficiently supporting large-scale batch data processing workloads with LLMs, building on DocETL.
• Makora – A platform that automatically writes, optimizes, and deploys high-performance GPU code, unlocking significant cost savings and deep infrastructure performance improvements across any hardware environment. (Fast Track Team)
• Project SWIFT (Hewlett Packard Enterprise) – Scalable, efficient and resilient AI runtime for infrastructure management. (Fast Track Team)
• UCSD-Systalyze (University of California, San Diego) – A full-stack, dynamic performance intelligence layer that optimizes GPU execution and sharing for superintelligence workloads.
• Systems for Future Intelligence Lab (University of Washington) – AI-driven infrastructure specialized to each model, workload, and hardware.
• TRASE (Rice University) – A unified inference stack that replaces the costliest operations in multi-turn AI agents (conversation history, attention memory, and tool lookup) with sub-linear approximations.

Benchmarking Team:

• MLCommons – Builds and deploys benchmarks that standardize AI measurement and enable informed decisions. As an engineering consortium supported by over 150 corporate, non-profit, and academic members, its staff and working groups create and maintain industry-standard assessments.

Teams have started their work in September of 2026; STRIDE will share insights into their work as teams reach new milestones. Follow along on STRIDE’s LinkedIn and via the STRIDE website: stride-ventures.com.

The AI Efficiency Challenge is administered by Start2 Group as the OT Contractor under an Other Transaction Agreement with the National Science Foundation.

About STRIDE Ventures – Managed & Operated by Start2 Group
STRIDE Ventures is a national platform for breakthrough innovation, launched by the U.S. National Science Foundation and Start2 Group. It supports NSF-funded bold, milestone-driven efforts to translate emerging technologies into real-world capabilities that advance U.S. competitiveness and resilience. By minimizing bureaucracy and maximizing momentum, STRIDE empowers teams to move fast, take risks, and deliver impact where it matters most.

About NSF TIP
The NSF Directorate for Technology, Innovation and Partnerships (NSF TIP) seeks to engage all Americans in accelerating critical and emerging technologies to advance U.S. competitiveness. The directorate partners across sectors to advance three strategies: accelerating critical and emerging technology, expanding the geography of American innovation, and building a competition-ready workforce. For more information about NSF TIP, visit nsf.gov/tip/latest.

Media Contact
Marie Klemme-Lacus
marie.klemme-lacus@start2.group