Research Teams at Leading Institutions Choose B3IQ to Own, Run, and Monetize Their AI Stack

NEW YORK, NY, August 11th, 2026, FinanceWire

Demand for private inference is surging as a way to control token costs and keep data private:

• Top research at NYU, Stanford, Dartmouth, and UH Mānoa are using B3IQ to power their AI work, including for cancer research and training specialized AI models.
• A 2026 Broadcom survey of 1,800 IT leaders found 56% of enterprises now run or plan to run production inference on private cloud infrastructure, while public-cloud use for the same workloads fell from 56% to 41% in a single year.
• Currently operating from a 27,000-square-foot Oregon facility, B3IQ plans to scale its inventory of U.S.-assembled NVIDIA GPU systems quickly to meet the growing demand.

Privacy is a growing problem in AI. Every prompt sent to a centralized AI provider is a deposit into someone else's vault: research data, business logic and ideas flow into systems the sender doesn't control, on terms they didn't set. The obvious fix, running AI on owned hardware, has been out of reach for nearly everyone. Why? High-end GPU systems cost tens of thousands of dollars, supply is constrained, and operating these systems require space and expertise.

Today, B3 Labs launched B3IQ to close that gap: a first-of-its-kind AI infrastructure provider designed to give universities, enterprises, and advanced users greater control over the hardware, models, and data underlying their AI workloads.

Until now, organizations seeking AI compute have chosen between two imperfect options. Renting from cloud providers means variable pricing, no ownership, and rationed access during GPU shortages. Buying outright means absorbing the full upfront cost along with the power, cooling and maintenance burden. B3IQ offers an alternative model, combining elements of both: ownership economics with hosted operations. 

B3IQ allows participants to purchase dedicated NVIDIA GPU systems, manufactured in the U.S. by B3 Lab’s portfolio company, Andromeda, and hosted in Oregon, through an incremental payment plan rather than full upfront costs. Through the B3IQ dashboard, owners can monetize unused capacity by matching it with compute demand, with revenue applied toward their hardware balance or kept as income. Upon full payment, owners retain the option to continue B3IQ hosting or take physical delivery of their hardware.

B3IQ's early users include faculty members, AI researchers and student-led teams at New York University, Dartmouth College, the UH Manoa, and Stanford University for whom GPU scarcity pose particular budgeting problems:

"B3IQ's owner-controlled model is a promising path between renting and buying — and that is why we decided to partner with B3IQ. Grants are fixed and awarded upfront; cloud costs are variable and can quietly consume a line item mid-project. Turning compute into a known, budgetable cost makes it easier to plan — and to answer to a PI or a grants office," said Pavel Bushuyeu, AI researcher at the University of Hawaiʻi. "Owning capacity also insulates us from scarcity. When GPU availability is tight, centralized providers ration access, and academic users tend to be deprioritized behind paying enterprise customers. With our own node, we're not competing for a slot when we need to run something. And when the system isn't in use for our work, the option to apply a share of revenue from its unused capacity toward the cost of the system helps offset the investment."

A significant market exists for researchers and organizations that cannot send sensitive data to third-party model providers like Anthropic or OpenAI. For instance, B3IQ pilot users such as Pavel Bushuyeu from the University of Hawaiʻi utilize the platform to run proprietary models for cancer research and robotics training, handling sensitive data that simply cannot be shared externally. AI providers also restrict keywords and topics outright, which can make entire research areas impossible: master's students in Professor Yorke E. Rhodes III's Ethical Tech CoLab at the NYU Center for Global Affairs build frameworks and simulations from war-zone evacuation data and model nation-level diplomatic negotiations, work that trips commercial content filters and has to run on infrastructure the team owns.

The company points to the rapid improvement of open-weight AI models as the reason the timing works: organizations can now run advanced workloads without relying exclusively on centralized model providers, provided they have private infrastructure to run them on. 

"Organizations want more control over where their AI runs, how their data is handled, and what they pay for compute," said Sean Geng, CTO of B3 Labs. "B3IQ brings those decisions into one system: dedicated hardware for private workloads and an opt-in network that puts unused GPU capacity to productive use."

Information on acquiring a GPU, contributing capacity or accessing compute through the network is available at b3iq.org.

About B3 Labs

B3 Labs builds software and hardware for enterprise AI. Founded in 2024 by a team of Coinbase alumni, the company has raised over $21 million from investors including Pantera Capital and Coinbase Ventures. B3 Labs operates two product lines: B3OS, an AI execution engine powering deterministic execution by agents within enterprise systems, and B3IQ, US-based GPU infrastructure that customers own outright. Learn more at b3os.org and b3iq.org.

SOURCE: B3 Labs