UCB RFQ 003970-Apr2026 - GPU Servers for Advanced Bioimaging Center

This is an OVERVIEW ONLY. This is not the entire RFP Document

Please accept mandatory guidelines in CalUsource to view event.

University of California, Berkeley

Advanced Bioimaging Center

RFx #003970-Apr2026

GPU Servers for Advanced Bioimaging Center

RFx Timeline – Listed below are the key action dates/times for this RFx the amount of time listed in the table are suggestions. Discuss with your Buyer on a realistic RFX timeline. The UC reserves the right to modify this timeline to best meet the needs of the University.

1. INTRODUCTION

The University of California, Berkeley (UCB) invites proposals for GPU servers for machine learning research on behalf of Advanced Bioimaging Center (ABC) within the Department of Molecular and Cell Biology.

University of California, Berkeley. The University of California, Berkeley is internationally renowned for excellence and pioneering achievements across all disciplines. The University of California was chartered in 1868 with Berkeley as its flagship campus. Today the world’s premier public university and a wellspring of innovation, Berkeley occupies a 1,232-acre campus with a sylvan 178-acre central core. From this home, Berkeley is committed to community engagement, public service, and industry partnerships in order to advance the economic and social well-being of our region, state, nation, and the world.

Department Background. The ABC is administered by the MCB Department and directed by MCB faculty member Eric Betzig and MCB faculty member Gokul Upadhyayula. This facility hosts two next-generation MOSAIC (Multimodal Optical Scope with Adaptive Imaging Correction) microscopes that combine several imaging modalities, including Lattice Light Sheet Microscopy with adaptive optics for fast, high-resolution, live-cell imaging. The multi-functional microscope has been designed to seamlessly switch between several operational modes—all capable of leveraging adaptive optics as needed for the biological sample. In addition to the two MOSAICs, the ABC is currently building two inverted adaptive optical structured illumination microscopes (iSOAR) to image dynamic subcellular processes across large volumes of living tissue in transparent organisms at an unprecedented combination of speed (∼10 TB/hr) and resolution (∼130 x 130 x 450 nm). This would enable studies ranging from in vivo single molecule kinetics, to subcellular organelle interactions, to the collective behavior of migrating cells during development.

ABC houses the computing infrastructure needed to handle and process the large data sets generated by adaptive optical lattice light sheet microscopy mode of the MOSAIC. ABC has an expandable cluster of 36 nodes, with a total of 1000 CPU cores 40 GPUs. Each node has 512 GB memory for handling and analyzing large 3D volumetric time series; four nodes are equipped with three high performance CUDA-enabled GPU cards suited for inferencing machine learning algorithms. The ABC microscopes are inter-connected to a 1000 TB ultrafast data storage server, a 600 TB fast solid-state drive server, 3000TB archival server, and linked directly to the computing cluster with high-speed fiber optics (100 Gbps) for real-time network data streaming and transfers.

2. PURPOSE

The purpose of this Request for Quote (“RFQ”) is to invite qualified bidders to prepare and submit proposals to furnish four (4) Rackmountable HGX B200 GPU Servers for machine learning research to the University of California, Berkeley (“UC Berkeley” or “UC”), in accordance with the requirements set forth in this RFQ.

3. MINIMUM / MANDATORY REQUIREMENTS

Bidders shall be eliminated from further consideration if Minimum / Mandatory Requirements are not met.

a) Business Experience: Proven experience serving public agencies or universities, including compliance with higher education policies.

b) Implementation Understanding: Clear articulation of the project scope, timeline, and alignment with university objectives.

c) Geographical Support: Proven ability to service hardware on the Berkeley campus. Non-local/out-of-state vendors must provide a detailed on-site support plan.

d) Expected Delivery Time: 8 to 16 weeks of UC Purchase Order received.

e) Warranty. 3 Year Return to Depot Warranty on Systems and Advanced Replacement Warranty on parts (except on GPU trays / NVIDIA parts), with labor.

4. EVALUATION CRITERIA

The evaluation process for this RFx will be made based on University of California Policy BUS-43, which is posted online. Factors that will be used to evaluate information include, but are not limited to:

a) Specification Adherence: Exact match (or demonstrably superior) to the stated technical specifications.

b) Pricing: Total cost of the proposed solution.

c) Delivery: Verified ability to deliver within the 16-week window (pricing must reflect this timeline).

d) Support & Warranty: Clarity of the 3-year support plan, including explicit details on which hardware elements are ineligible for advanced parts replacement.

e) Customer Service: Availability and quality of support channels (phone, email, live chat, self-service).

f) Sustainability: All suppliers to the University of California are required to meet all applicable sustainability standards and requirements as outlined in the UC Sustainable Practices Policy (https://policy.ucop.edu/doc/3100155). See the University of California Sustainable Practices Policy (https://policy.ucop.edu/doc/3100155) and the UC Berkeley Sustainability Office (https://sustainability.berkeley.edu/home) information sites for more detail.

g) UC Terms: Supplier will agree to the UC Terms and Conditions of Purchase, UC Agreement Templates and any applicable UC Appendices associated with an awarded contract. See Section 7 and Exhibit 4.

h) References: UC reserves the right to check references other than those provided by the Bidder, and to ask follow up questions to clarify any responses received. Bidders should provide references for projects of a similar size and scope and similar institutions to UC Berkeley.

5. EVALUATION PROCESS

Bids will be evaluated based on the criteria listed in the RFx in 2 or 3 tiers based on the need for a Bidder Interview. A complete bid response contains answering all questions in Section 7 of this RFP, proving a response to all pricing in Exhibit 1, and returning applicable Exhibits.

a) Tier 1 - Minimum / Mandatory Requirements: Bidders shall be eliminated from further consideration if Section 4: Prequalification – Minimum / Mandatory Requirements aren’t met.

b) Tier 2 - Bidder’s Proposal Evaluation: Evaluation criteria will be based on responses to the RFx requirements.

c) Tier 3 - Optional requests for additional information or virtual presentations/interviews, at the discretion of the UC review team. Clarifying information from this tier may be considered when evaluating the proposals. The UC reserves the right to eliminate this tier if it determines the written bids sufficiently address the evaluation criteria.

d) Contacting Phase - Notice may be given to the top ranked Bidders. UC will notify the Apparent Awardee(s) via an Intent to Award notice, request best & final offers and move to acceptance and contract execution activities. UC reserves the right to reject all proposals, make more than one award, or make no award. Based on the results of this RFx process, UC reserves the right to request one or more separate proposals for selected items, with a select group of suppliers, including direct manufacturers of these products. Based on the results of such requests, UC or its locations may enter into contracts with other suppliers to provide those products, while remaining UC locations may buy those products through this RFx. Any contract awarded pursuant to this RFx will incorporate the RFx requirements and specifications, as well the contents of the Proposal as accepted by UC, and will be in writing.