FedGeoDay 2026: AI, Economics, and the Government's Push for Data Resilience
Watch the full FedGeoDay coverage by Clicking on the image
Walking the sprawling halls of the United States Census Bureau headquarters for FedGeoDay 2026, the atmosphere was a unique, often contradictory blend of technological optimism and pragmatic survivalism. Co-chaired by Monica Mardell and Jackie Kazil, the two-day event brought together the brightest minds across the federal geospatial ecosystem, private industry, and open-source communities. But beneath the familiar hum of networking, vendor pitches, and platform demos, a deeper, more urgent narrative was taking shape: the vital necessity of data resilience in an era defined by constrained budgets and the rapid, sometimes chaotic acceleration of the artificial intelligence revolution.
Andrea Johnson, Acting Division Chief of the US Census Bureau’s Geography Division, captured the prevailing sentiment perfectly on the morning of Day 2. She invoked the National Geographic definition of an ecosystem, "a dynamic functional unit consisting of a specific geographic area where living organisms interact with each other and their non-living environment", and mapped it directly onto the federal geospatial community. This is an ecosystem under severe stress, and as speakers noted time and again, collaboration is the only way through.
To address this need for collaborative problem-solving, Kazil introduced an "unconference" format for the afternoon of Day 2—a grassroots, dynamic scheduling approach where attendees literally built the agenda on a whiteboard using sticky notes and markers. It was a fitting metaphor for the conference as a whole: moving away from rigid, top-down structures toward flexible, community-driven solutions.
Here are the major highlights from FedGeoDay 2026, and what they reveal about the complex, evolving future of federal data stewardship.
The AI Revolution: Ontologies, Agents, and the Scraper Threat
Unsurprisingly, Artificial Intelligence dominated much of the conversation. However, unlike the breathless hype often found at broader tech summits, the geospatial community is looking past the novelty of generative AI toward the thorny, practical realities of integration, semantic structures, and the massive infrastructural toll AI is taking on open data.
In one of the most provocative moments of Day 1, Andrew Turner challenged the industry's traditional obsession with rigorous data standardization. "I'm going to say something radical here for a moment amongst this crew," he noted, acknowledging the heavy presence of standards-focused architects in the room. "I think we're at the end of standards. I don't think they'll matter anymore... AI models have been around for a while, you can now have it talk to any API, and it can introspect that and figure it out."
This concept, that AI can dynamically interpret and map disparate datasets on the fly without requiring them to adhere to a rigid Open Geospatial Consortium (OGC) standard, promises unprecedented speed in data integration. Jerry Johnston of Deloitte echoed this potential, introducing Earth Observation (EO) foundation models like the open-source "Clay" model. By encoding 70 million satellite images into dense vector embeddings, these models allow analysts to instantly detect subtle surface deformations or map changes along international borders at a fraction of traditional computational costs.
However, the delegation of geospatial reasoning to LLMs brings up massive concerns regarding attribution, hallucinations, and system abuse. Matt "Hutch" Hutchinson of Woolpert countered the "end of standards" argument by highlighting that probabilistic systems (LLMs) cannot blindly orchestrate deterministic geospatial tools without disastrous results. He argued for OWL-based ontologies, shared semantic frameworks, to provide strict guardrails for AI agents, ensuring they don't hallucinate analytical workflows.
Furthermore, the physical cost of AI on the data ecosystem is becoming a crisis. Maggie Cawley of OpenStreetMap US (OSM US) delivered a stark warning: AI scrapers are wreaking havoc on open-source infrastructure. Scrapers are causing regular outages on osm.org, trampling the Overpass API, and driving up cloud costs, all while the AI industry gives little to nothing back to the maintainers. Katie Baynes, NASA's Earth Data Officer, echoed this alarm, revealing that NASA's data egress has roughly quadrupled as rogue AI agents (like hastily written Claude Code snippets) attempt to pull massive, inefficient queries across NASA's 150-petabyte archive.
As one panelist highlighted, the ultimate goal is seamless but strictly attributed integration. "If we do it right, people can go into their AI assistant of choice and say, “Tell me what NASA knows about sea level change”... I'm hopeful that the recognition that the data sources…NASA, DOI, USGS…it's not OpenAI that's creating the answer. It's the data that we're all producing."
The Data Crisis: Rescue Operations and the Power of Storytelling
This tension surrounding AI feeds directly into a broader, government-wide mandate: Data Resilience. How does the government maintain critical public data infrastructure when federal funding is volatile and political priorities shift?
Joel Gurin of the Center for Open Data Enterprise outlined what he called "America's Data Crisis," noting the quiet disappearance of vital datasets over the last year, including the halting of the USDA food security survey and the systematic removal of SOGI (sexual orientation and gender identity) data from federal records.
When official channels falter, the burden falls on the community. Frank Donnelly of Brown University detailed a heroic, volunteer-driven "Data Rescue Project" that spent the summer and fall archiving the DHS HIFLD Open dataset before it was deprecated. The team managed to catalog over 400 critical infrastructure layers, moving them to Data Lumos at the University of Michigan to ensure researchers didn't lose access overnight.
Former US Chief Data Scientist Denice Ross delivered a powerful keynote arguing that the geospatial community's best defense against defunding is radically improving how it tells the story of data to the American public. She laid out a "baker's dozen" of critical federal datasets with tangible human impacts: the USGS North American Bat Monitoring Database, which streamlines wind farm permitting while tracking bats that provide $53 billion a year in free pest control; NOAA NEXRAD radar data, which pilots actively use to avoid bird strikes; and the USFS National Fire Danger Rating System, which powers the iconic wooden signs at national park entrances.
"The public safety community has to be right," noted Jesse Osborne during a lightning talk on “LINK”, a new AI-powered tool for local infrastructure network knowledge. When a wildfire takes out a water pump, local emergency managers need to know exactly which hospitals are downstream. Resilient data saves lives, but only if the public understands its origin.
Budget Constraints and the "I Will Survive" Mentality
If AI and data rescue were the thematic anchors of the conference, budget reality was the operational one. During a Day 2 presentation reviewing the annual federal GIO survey, Jaymes Cloninger of Motivf painted a stark, sobering picture.
When asked about the adequacy of current federal funding, the overwhelming response shifted drastically from "somewhere in the middle" last year to definitively "inadequate" this year. When asked about top challenges, Cloninger noted the dominant write-in answer was simply: "budget, budget, salaries, salaries, and budget."
Yet, this financial pressure is acting as a potent catalyst for innovation. The survey revealed a significant spike in agencies actively exploring cost-sharing and inter-agency collaboration to sustain their programs. Isolated silos are becoming a luxury that agencies can no longer afford.
Interestingly, when the survey asked what theme song best represents their agency's geospatial program, the top answer shifted from Survivor's aggressive "Eye of the Tiger" to Gloria Gaynor's enduring anthem, "I Will Survive."
As part of this survival strategy, open-source tools have experienced a massive perception shift, transitioning from a risky alternative to mission-critical infrastructure. Robert Pitts of ActioNet emphasized the synergy of cloud infrastructure, open-source software, and AI as the only viable path forward for federal modernization, allowing agencies to bypass massive capital expenditures. Platforms like GeoServer 3 and Core Spatial (an AI-enabled open-source GIS platform built for secure federal deployment) are proving that agencies can achieve secure, accredited systems without crippling vendor lock-in.
"Place is the Tie That Binds": Bespoke GDP and the 1km Grid
Day 2 brought an exciting shift toward the intersection of geospatial data and macroeconomics. As Andrea Johnson reflected on the aftermath of Hurricane Katrina twenty years ago, she recalled the agonizing difficulty of trying to precisely identify which businesses and economic zones were impacted.
"Today, the Census Bureau has significant initiatives to identify ways in which we can link job data, economic data, people data, and business data all with place," Johnson stated. "Place is the tie that binds. Place is the beginning of that ecosystem."
This sentiment was enthusiastically echoed during an unprecedented "Between the Ferns" style fireside chat featuring the heads of three major statistical agencies: Ron Jarmin (Census), Dr. Vipin Arora (BEA), and Bill Wiatrowski (BLS). The consensus was clear: the public is demanding hyper-local economic insights. Mayors and city planners no longer just want national metrics; they want granular data.
Dr. Arora floated a fascinating concept: a "bespoke GDP", a user-defined regional economic measure that could be drawn on a map rather than constrained to arbitrary county lines. "What a service to taxpayers if we could do something like that," a BLS representative agreed. "But we're not going to be able to do it alone. We're going to need your help."
To achieve this, statistical agencies are fundamentally changing how they operate. Hector Ferronato detailed a massive Census initiative using commercial satellite imagery and machine learning to track "housing starts"—monitoring pixels as they transition from vegetation to excavation to foundation—to supplement traditional building permit surveys.
Furthermore, Census Geography's Josh Coutts made a massive announcement that sent ripples through the audience: the forthcoming release of a one-square-kilometer statistical grid covering the contiguous US, Alaska, and territories. Arriving later in 2026, this grid will finally free users from messy administrative boundaries, allowing direct integration of demographic estimates with Earth observation data and USGS 3DEP elevation products. Quoting Tobler's First Law—"everything is related to everything else, but near things are more related than distant things"—Coutts noted the grid is the foundational step toward mapping true localized economic and social realities.
Closing Thoughts…
FedGeoDay 2026 proved beyond a shadow of a doubt that the federal geospatial sector is in a period of intense, perhaps unprecedented, transition. The community is caught between the limitless potential of AI integration, the threat of infrastructure abuse, and the harsh, grounding realities of constrained federal budgets.
Yet, despite the financial anxiety and political volatility surrounding federal data, there is a distinct sense of optimism and agency. As the Day 2 GIO survey humorously pointed out, when asked what fictional character best represents their role, the community shifted their collective answer from "MacGyver", frantically trying to escape a burning building with a pocket knife and duct tape, to "Dora the Explorer."
It is a profound shift in self-perception. Dora is proactive, inquisitive, prepared, and fundamentally relies on a map to navigate the unknown. The federal geospatial community is no longer just patching together failing legacy systems in a panic; they are deliberately mapping out new territory. By utilizing ontologies to tame AI agents, rescuing endangered datasets through sheer community willpower, and revolutionizing how we map the national economy with grids and satellites, the professionals at FedGeoDay are ensuring that the nation's spatial infrastructure remains not just functional, but profoundly resilient.
