Reclaiming the Stack: Why Researchers and Nonprofits Are Building Their Own Digital Homes
Somewhere in the terms of service of nearly every major cloud platform is a clause that academic institutions and nonprofit organizations have learned to read very carefully. The language varies, but the underlying reality is consistent: when you store your data on someone else's infrastructure, you are operating under their rules, subject to their pricing decisions, and dependent on their continued interest in serving your sector.
For years, the convenience and apparent cost-effectiveness of platforms like Google Workspace, Microsoft Azure, and Amazon Web Services made that dependency feel acceptable — even rational. The calculus is changing. Across the United States, a growing coalition of researchers, librarians, archivists, and nonprofit technologists is reaching the same conclusion: the long-term costs of platform dependence, measured not just in dollars but in autonomy, privacy, and institutional resilience, are higher than they initially appeared.
The Sovereignty Problem
The concept of data sovereignty — the idea that an institution should control where its data lives, who can access it, and under what legal jurisdiction it resides — has moved from a theoretical concern to a practical one with real consequences.
Consider the situation facing research institutions that work with sensitive human subjects data. Federal regulations under HIPAA and IRB frameworks impose strict requirements on how such data is stored and who may access it. When that data lives on a commercial platform, compliance depends on the platform's cooperation, its infrastructure decisions, and its response to third-party legal requests. Several high-profile incidents over the past decade — including cases where commercial platforms received government subpoenas for data held on behalf of academic clients — have concentrated institutional minds.
"The moment we realized our IRB-approved dataset could be subject to a legal process we had no standing to contest, the conversation about infrastructure changed completely," said one data science director at a mid-sized research university in the Midwest, who spoke on background because their institution's cloud contracts are still under negotiation. "We were not in control of our own research."
This is not an abstract concern. The American research enterprise depends on the trust of human subjects, the integrity of longitudinal datasets, and the confidence of international collaborators — all of which can be compromised when data governance is effectively outsourced to a commercial third party.
The Economics of Independence
Beyond sovereignty, the economic argument for community-owned infrastructure has become substantially more compelling in recent years. Cloud platform pricing, particularly for storage-intensive research workloads, has increased steadily. Egress fees — charges for moving data out of a platform — have emerged as a particularly painful cost for research institutions that need to share large datasets with collaborators at other institutions.
By contrast, federated and community-owned infrastructure models allow institutions to pool capital expenditure, share operational expertise, and negotiate collectively for hardware and bandwidth. The Internet2 consortium, which provides advanced networking and cloud services to US research and education institutions, represents one mature example of this model. Member institutions collectively fund infrastructure that no single university could afford independently, while retaining far greater control over how that infrastructure is governed than any commercial arrangement would permit.
Smaller-scale examples are proliferating. Library consortia are deploying shared instances of open-source repository platforms like DSpace and Islandora. Independent research organizations are standing up federated identity systems using open standards rather than delegating authentication to commercial identity providers. The pattern is consistent: pooled resources, shared governance, and open-source software as the enabling technology.
What Research Freedom Looks Like in Practice
The freedom argument for community-owned infrastructure is perhaps the most difficult to quantify but the most important to articulate. When a researcher's primary data repository, collaboration tools, and publication infrastructure are all controlled by commercial entities, those entities acquire a subtle but real influence over what research is practical to conduct.
Platforms that make certain data formats easy to work with and others difficult, that provide excellent tooling for some research methods and poor tooling for others, or that deprecate features relied upon by specific research communities are not making neutral technical decisions. They are shaping the research environment in ways that serve their own product roadmaps rather than the needs of the scientific community.
Open infrastructure — software whose source code can be examined, modified, and extended by the institutions that use it — breaks this dependency. A research team that needs a specialized data ingestion pipeline can build one and contribute it back to the shared codebase. An institution with unusual archival requirements can adapt the platform to meet them. The research environment becomes an extension of the research community's own judgment rather than a constraint imposed from outside.
The Barriers That Remain
None of this is to suggest that the transition to community-owned infrastructure is straightforward. The challenges are real, and honesty about them is a prerequisite for making progress.
The most significant barrier is operational capacity. Commercial platforms succeed in part because they abstract away enormous operational complexity. Running your own infrastructure means hiring or developing staff who understand systems administration, security patching, backup management, and incident response. For institutions with small IT teams and limited budgets, this is a genuine obstacle — not an excuse for inaction, but a real constraint that must be addressed through shared services models and investment in training.
Funding is the second major barrier. The upfront capital costs of deploying community-owned infrastructure can be substantial, and the grant funding landscape for infrastructure projects remains underdeveloped relative to the funding available for research that uses infrastructure. The National Science Foundation has made progress through programs like the Open Storage Network and the ACCESS program (successor to XSEDE), but the gap between what is funded and what is needed remains wide.
Finally, there is the challenge of interoperability and standards. Community-owned infrastructure only delivers on its promise if different institutions' systems can communicate with one another. This requires sustained investment in open standards development — work that is technically demanding, often unglamorous, and poorly compensated relative to its importance.
What a Healthier Ecosystem Requires
The movement toward community-owned research infrastructure is not anti-commercial. Commercial platforms will continue to play a role in the research ecosystem, particularly for workloads where their scale and tooling genuinely serve research needs. What the movement is arguing for is a better balance — one in which institutions retain meaningful alternatives to commercial dependency and the open-source infrastructure commons is adequately funded and maintained.
Achieving that balance will require action on multiple fronts. Federal funding agencies need to treat infrastructure as a first-class research investment rather than an indirect cost. Academic institutions need to treat infrastructure expertise as a valued scholarly contribution rather than a support function. And the broader open-source community needs to continue developing the governance models and sustainability mechanisms that allow community-owned projects to endure beyond the tenure of their founding teams.
At Condor Platform, this challenge sits at the center of our mission. The tools, documentation, and community infrastructure we provide exist precisely to lower the barrier for research teams and nonprofit organizations that want to build and operate their own digital environments. The technical path is clearer than it has ever been. The harder work — cultural, financial, and political — is the work that remains.