timeline
title Key Developments Since the RDWG Roadmap
section Before the roadmap
Nov 2022 : ChatGPT launches — generative AI era begins
Dec 2022 : RDWG Roadmap published
: UC Research Data Policy takes effect
section Year 1
Jan 2023 : NIH Data Management & Sharing Policy takes effect
: Every NIH proposal now requires a funded DMP
2023 : RDWG Phase 2 — 33 campus leadership meetings confirm gaps
: Implementation task analysis and priority matrix completed
section Year 2
Jun 2024 : Google Drive 10PB migration deadline passes
2024 : RDWG Deans Council Brief finalized
section Year 3
Oct 2025 : UCLA Academic Senate delivers OneIT consultation responses
Dec 2025 : OSTP public access mandate takes full effect
section 2026
Mar 2026 : One IT Research Working Group charged by EVCP and CIO
What’s Changed Since December 2022
RDWG Roadmap Gaps and New Developments for the One IT Research Working Group
The RDWG roadmap is a strong evidence base, but it was completed in December 2022 — before several major developments reshaped the research technology landscape. This document identifies:
- What’s materially changed since the roadmap was written
- What the roadmap underweighted or missed at the time
- What this means for One IT WG recommendations
1. Generative AI — The Biggest Gap
The roadmap was finalized the same month ChatGPT launched (November 2022). Generative AI was not on the map. Since then, it has become one of the most significant forces reshaping research across every discipline.
What the roadmap missed:
GPU infrastructure demand has exploded. The roadmap recommended GPU node expansion for Hoffman2, but the scale of demand for AI/ML compute — now across the humanities, social sciences, and life sciences, not just CS — vastly exceeds what was anticipated. Research institutions are now competing for GPU allocations at a level the 2022 document could not have foreseen.
Institutional data and AI services. Researchers are using commercial AI tools (ChatGPT, Copilot, Gemini) with institutional and research data. The compliance, IP, and data governance implications are largely unaddressed at UCLA. Who owns outputs generated using AI tools with UCLA data? What can be sent to external AI providers? These questions land directly on research IT and data governance — exactly the ORDR’s mandate.
Research integrity and reproducibility. AI-generated text, code, and analysis in research outputs creates new reproducibility challenges. Journals, funders, and the NAS are actively developing standards. UCLA needs a position and infrastructure response.
Compute as a research equity issue. Access to GPU compute is now a determinant of research competitiveness across disciplines. Labs with resources buy their own; others can’t compete. This is a new version of the same inequity problem the roadmap identified for storage — and it argues for the same centralized solution.
Research data for AI training. UCLA has significant unique datasets (archival collections, clinical data, specialized corpora) that could be used for AI research. Governance frameworks for this use case don’t exist.
Important
Implication for One IT: The WG should explicitly address AI research infrastructure — GPU compute access, institutional AI service governance, and data policies for AI use. The RDWG framework provides the right structure; the content needs updating for the AI era.
2. The Compliance Environment Has Intensified
The roadmap anticipated increasing funder mandates but didn’t have the full picture:
“NIH must know that we don’t have the capacity to do this. We’ve been in conversations with our program officers. They ask us what campus resources we have, and we tell them we’re responsible to do it for the grant. Our program officers are surprised that we don’t have any support in this.”
— UCLA faculty member, RDWG campus sessions (2023)
NIH Data Management and Sharing Policy took effect January 25, 2023 — six weeks after the roadmap was published. Every NIH proposal now requires a funded data management and sharing plan. UCLA still lacks the coordinated infrastructure (ORDR, service catalog, compliance lead) the roadmap said was essential to respond to this.
OSTP “Nelson Memo” public access mandate took full effect for agencies by December 2025. Federally-funded research must be publicly available immediately at no cost. This affects publications AND research data. UCLA’s position for compliance is unclear.
Export controls on AI and semiconductor research have intensified significantly. Research involving controlled technologies, dual-use data, and foreign national researchers is under increasing scrutiny. P3/P4 compliant environments are no longer just a nice-to-have for health data — they’re required for a growing share of federally-funded research. The roadmap identified this gap; it has grown substantially.
UC Research Data Policy (effective July 2022, noted in the roadmap) is still not fully operationalized campus-wide. The compliance lead the ORDR was supposed to provide still doesn’t exist as a clear role.
A P3/P4 answer already exists — and is already institutional: UCLA Library operates UCLA Redivis, a production-ready sensitive data platform in active institutional use. It supports granular permissions, Data Use Agreement (DUA) management, SSO, DOI minting for datasets, and Python/R APIs for researcher workflows. Its multi-organization architecture means the institutional framework is already established — the Library is an active organization within it, and the buy-in model is in place. Any WG recommendation on sensitive data should build on this existing infrastructure, not duplicate it.
Important
Implication for One IT: Compliance is now more urgent and more complex than the 2022 analysis captured. The ORDR role and the P3/P4 enclave are now critical infrastructure, not aspirational goals. For the P3/P4 enclave specifically, UCLA Redivis already provides the core platform — the missing piece is institutional coordination and VCR-level governance, not a new build.
3. The Google Drive Crisis Landed
“Storage is a huge issue. Chairs would email me about storage — backup storage, collaboration space. They’re putting data all over the place. The lack of a clear set of choices means everyone is finding their own way, so keeping this secure is impossible.”
— UCLA associate dean for research, RDWG campus sessions (2023)
The roadmap flagged a specific, dated crisis: approximately 10PB of UCLA research data stored in Google Drive in excess of the new Google storage limits, with a June 2024 deadline to migrate or pay $1.8M/year in overage fees.
That deadline has now passed — and UCLA’s official response is documented.
What actually happened at UCLA:
According to UCLA Digital & Technology Solutions, the campus undertook a major multi-year effort to contain and reduce Google Workspace storage. The result: UCLA reduced campus Google Workspace data storage by over 14PB — far exceeding the ~10PB the roadmap had flagged — “achieving quota compliance on time and avoiding any disruptions or additional costs to campus.” (UCLA DTS, Google Workspace Service Adjustments)
That is the answer to “what happened”: 14PB of data — research files, datasets, shared drives accumulated over years because no alternative centralized storage existed — had to be deleted or migrated. The effort took over two years of active campus engagement.
The crisis is not resolved. As of March 2026, a temporary quota extension is still expiring — the permanent compliance deadline is April 7, 2026. Accounts exceeding their quota after that date will experience service interruptions. UCLA is now moving toward a paid storage model: a pilot purchase program launches summer 2026 (no cost), with fees applying from January 2027 onward.
The practical outcome: instead of the Bruin Research Data Hub — a centralized research storage baseline proposed in 2022 — UCLA faculty will pay for commercial Google storage starting in 2027.
What the peer institution pattern shows:
At the University of Texas at Austin, the scale was comparable: 6.4 PB forced into a 545 TB quota, affecting 270,000 accounts — a 90% reduction. The pattern held across higher education: when institutions lack centralized research storage, researchers fill whatever vessel exists. Underfunded labs store datasets in Google Drive not because it is the right tool but because it is the only centrally-provided option. When that option disappears, the crisis it reveals was already there — just invisible. (EdTech Magazine, “How Higher Ed Institutions Are Responding to Google Storage Limits,” January 2023)
What good centralized infrastructure looks like:
Stanford’s Sherlock HPC cluster — available at no cost to all Stanford PIs — provides 75,000 CPU cores, 1,200+ GPUs, and 77 PB of long-term research data storage. Stanford faculty do not store research data in Google Drive because they have a proper home for it. The Bruin Research Data Hub was proposed precisely to provide UCLA with an equivalent baseline.
Important
Implication for One IT: The Google Drive crisis was not a one-time event — it was the predictable result of having no centralized research storage baseline. The campus response — deleting 14PB and moving toward paid commercial storage — was expensive, disruptive, and left researchers without a research-grade alternative. The Bruin Research Data Hub would have prevented this. This WG is deliberating while the April 7, 2026 compliance deadline is active.
4. The ORDR Was Never Created
“I am always stunned at the 17 places where the same things are happening and encourage you to make sure that those overlaps are foregrounded and addressed.”
— UCLA dean, RDWG campus sessions (2023)
The roadmap’s first and most foundational recommendation was creating the Office of the Research Data Rockstar (ORDR) — a central coordination unit with clear ownership and accountability for research data campus-wide.
As of 2026, this role does not appear to exist in its proposed form. The result, as the roadmap predicted, is exactly the “leaderless vacuum” scenario: multiple units respond inconsistently (or not at all) to campus-wide research data issues, with no clear owner.
Important
Implication for One IT: The absence of the ORDR is a documented, self-fulfilling problem. The WG’s governance recommendations should directly address this gap. The ORDR concept maps cleanly onto what One IT needs in a research data coordination function.
5. Research Software and Code — Underweighted in 2022
The roadmap mentioned code repositories and software tools but treated research software primarily as a storage and sharing problem. Since 2022:
- Research software engineering (RSE) has emerged as a recognized professional role. Major institutions (Michigan, Stanford, NYU) now have dedicated RSE teams. UCLA does not have a coherent RSE capacity.
- FAIR for software (FAIR4RS principles) are now funder expectations alongside FAIR for data.
- Code reproducibility has become a major journal and funder requirement. Researchers need support writing reproducible, shareable code — not just storing it.
- GitHub/GitLab institutional instances are now common at peer institutions, providing version control, collaboration, and archiving in one place.
Tip
Implication for One IT: Research software support belongs in the research IT portfolio, not generic IT. This argues for specialist staff (RSEs) embedded in or closely coordinated with research units — not consolidated into a campus IT help desk model.
6. Cloud for Research — More Complicated Than Anticipated
The roadmap called for a managed Cloud for Research program (AWS, GCP, Azure). The reality at institutions that have implemented this is more complex:
- Cloud costs for research computing are unpredictable and often shocking. GPU instances especially. Researchers frequently exceed expected costs or avoid cloud because the billing model is opaque.
- Skills gap is significant. Researchers need training and support to use cloud effectively — not just access.
- Data transfer costs (egress fees) create lock-in and unexpected expenses.
- Security and compliance in cloud is more complex than on-premises in many ways.
Several peer institutions have pulled back from aggressive “cloud-first” research computing positions and moved toward hybrid models with on-premises capacity for predictable workloads and cloud for burst/specialized needs.
Tip
Implication for One IT: Cloud for research is essential but requires more sophisticated support than the 2022 roadmap described. Cost management, training, and a hybrid model are now clearly the right approach.
7. Data Sovereignty and Equity Frameworks
The roadmap briefly mentioned CARE principles (Collective benefit, Authority to control, Responsibility, Ethics — addressing Indigenous data governance) but did not develop this into a full recommendation.
Since 2022:
- Indigenous data sovereignty has become an active area of policy development at UC and nationally
- Community-based participatory research data governance is increasingly required by funders for health equity research
- Data equity — who has access to data, who benefits from data about them — is an emerging compliance and ethics concern
Tip
Implication for One IT: Data governance frameworks need to explicitly address these dimensions, particularly for health-related and community-engaged research.
8. DataX — What Happened?
The roadmap positioned DataX as a key partner and driver — the $10M initiative was expected to create demand for research data infrastructure and provide a pilot use case for the Bruin Data Hub.
DataX had a three-year initial funding window. Its status and trajectory as of 2026 is worth understanding:
- Did it generate the expected demand for research data infrastructure?
- Did the infrastructure investments DataX needed ever materialize?
- What does DataX leadership say about research data infrastructure gaps?
Tip
Implication for One IT: DataX was supposed to be both a demand signal and an early adopter for RDWG proposals. Understanding what actually happened is important context for the WG.
9. The 2023 Campus Validation: Gaps Confirmed, Action Stalled
Following the roadmap’s publication, the RDWG conducted an extensive review campaign: 33 meetings with campus leadership — deans, associate deans for research, chairs, and IT directors from nearly every school — plus four public faculty forums and sessions with Academic Senate committees (CDITP, COR, COLASC), and individual meetings with EVCP Darnell Hunt and Deans Council. Notes were independently coded by four reviewers.
The findings were unambiguous. Participants were strongly supportive and described the challenges articulated in the roadmap in their own terms. The quotes below are drawn directly from those sessions.
Faculty and Leadership — In Their Own Words
Grants and research leaving UCLA
“We’re thinking about the follow up to this grant, and we’re saying let’s do the next one at Notre Dame.”
— UCLA faculty executive committee chair
Faculty are actively considering routing grants through collaborating institutions. Researchers are already relying on services from UC Davis Health Cloud Innovation Center, Wharton Research Data Services, San Diego Supercomputing Center, and the University of Chicago’s OCHRE environment.
Recruitment and retention
“We’ve got faculty coming from other institutions and they say, ‘Hey, I’m coming from X and it was so easy there.’ It feels like they’ve slid backwards in terms of their storage needs.”
— Associate dean for research
Recruitment and retention came up repeatedly across schools. A lack of research infrastructure was cited as the reason a star recruitment failed.
Money lost, funders frustrated
“We’re putting out fires every day. There’s a lot of money being lost as well. Delays. NIH gets upset. Other institutions pull back funding. [It] becomes very real both for the PI and the School.”
— Principal investigator
One PI had to shift staff from research to data management solely to meet funder requirements for publicly available datasets.
Duplicative effort across the campus
“I am always stunned at the 17 places where the same things are happening and encourage you to make sure that those overlaps are foregrounded and addressed.”
— Dean
Faculty and units are independently assembling storage, computing environments, and data sharing infrastructure — spending grant money and time on work that should be centrally provided.
Secure storage — a gap with real consequences
“We can’t currently provide HIPAA-compliant storage. If we can’t help them, they’ll roll their own and it’s not secure and not scalable.”
— UCLA IT director
“We can’t currently provide HIPAA-compliant storage. We’re hoping for help for folks with a foot in the med center that work with human subjects. If we can’t help them, they’ll roll their own and it’s not secure and not scalable.”
— IT director
“Storage is a huge issue. Chairs would email me about storage — backup storage, collaboration space. … Establishing that baseline is low-hanging fruit. Everyone needs storage space: Box, Google Drive, Synology, research data backup, co-storage, allocations [with] permissions.”
— Associate dean for research
The bureaucracy is blocking research
“It feels like there are policies in place, but there’s an inflexibility to take something case-by-case and it requires eight people to find a solution. It should be easier than that.”
— Faculty member
“NIH awards are made to UCLA and not UC Health, but UC Health requires nonsensical oversight that, more often than not, is not applicable to non-human data collection or non-patient behavioral science data.”
— Faculty member
Data services — capacity simply doesn’t exist
“NIH must know that we don’t have the capacity to do this. We’ve been in conversations with our program officers. They ask us what campus resources we have, [and] we tell them we’re responsible to do it for the grant. … Our program officers are surprised that we don’t have any support in this.”
— Faculty member
What the Campus Prioritized
When mapped to the roadmap, campus feedback identified specific HIGH-priority items — all requiring significant funding and dedicated staff that was never allocated:
- Hire the Research Data Rockstar (ORDR) — executive decision required, national search needed
- Cohesive research service catalog with web presence — identified as ideally owned by ORDR; if none exists, where does it go?
- Expand DSC data-related training — capacity exists but is insufficient
- Hire specialized data scientists — some capacity exists, simply not enough
- P3/P4 compliant HPC/storage environment — IT directors confirmed they cannot currently meet this need
- Research data backup solutions — requires a decision, funding, and cross-unit agreement
- Storage baseline (Bruin Research Data Hub) — described by deans as “low-hanging fruit” that would immediately reduce fragmentation
The roadmap update concluded that moving forward required a new charge letter from the VCRCA and EVCP, implementation teams, and Budget & Finance commitments. None of these steps were completed before the effort stalled.
Important
Implication for One IT: The 2023 validation campaign is not background reading — it is direct evidence. UCLA’s academic leadership explicitly confirmed the gaps, named the priorities, and expected action. What the One IT WG is now analyzing is not a new discovery: it is a two-year-old mandate that was never funded. The WG’s recommendations should carry that weight.
Source: RDWG Roadmap Update and Moving Forward, August 7, 2023