The Scoreboard

The Race tracks one central question:

Is AI concentrating power or dispersing capability?

This page applies that question to current events and real-world examples. It is the observation layer of the site: short, grounded, and meant to show how the framework appears in public decisions, market moves, infrastructure projects, workplace changes, policy fights, and institutional choices.

Current Reading - May 2026

Mixed, but more contested than expected.

Concentration remains the default path, especially as frontier AI firms move deeper into deployment and organizational redesign. But May also showed a real counter-pattern: state and local governments are beginning to govern AI’s physical and economic footprint through moratoria, utility challenges, tax-incentive reviews, workforce planning, and public engagement.

Bottom line: Concentration still has the stronger structural position. But civic infrastructure is beginning to appear.

The full event log with descriptions of each event and links for more info is located here.

Interpretation

May’s strongest theme was not model capability. It was governance catch-up.

The public conversation around AI often treats the race as a contest among models, chips, companies, and national governments. But this month’s signals point to another arena: whether ordinary public institutions can develop the capacity to govern AI’s consequences before those consequences become hard to reverse.

The clearest example is compute infrastructure. Data centers are becoming local political objects. They require land, power, water, grid investment, diesel backup, tax incentives, utility coordination, and local approval. That makes them visible in a way that software often is not. In May, multiple jurisdictions used moratoria, permit pauses, phased approvals, utility complaints, and tax-exemption reviews to slow the default buildout path.

That matters because civic infrastructure often begins with time. A pause is not a solution. But it can create the public space needed for hearings, standards, disclosure, utility review, environmental analysis, community-benefit discussions, and political accountability.

The other major theme was transition shock. California’s executive order on AI workforce disruption and benefit-sharing treated AI labor-market change as a public problem, not merely a private-firm efficiency story. Utah’s AI Workforce Credential similarly points toward public education as a distribution channel for AI capability.

At the same time, the concentration signals were serious. OpenAI’s Deployment Company suggests that the next concentration layer may not be just models or compute, but implementation capacity: the ability to redesign institutions around AI. Groupon’s AI-native restructuring points to another possible concentration dynamic: firms using AI to reduce headcount and rebuild workflows around leaner organizations.

The main lesson from May:

The race is moving from abstract AI capability into institutional terrain: land use, utilities, labor markets, workforce systems, public engagement, tax incentives, and organizational redesign. That is where concentration or dispersion will increasingly be decided.

Scored Signal Clusters

1. Compute infrastructure became governable

Direction: Dispersion-leaning
Confidence: Medium-high

The strongest May pattern was the spread of public governance actions around data centers. Denver approved a one-year moratorium. Reno paused new applications. Hill County, Harlingen, Jefferson County, and Huron County adopted moratoria or pauses. Florida enacted state guardrails around water and electric-service costs. Ohio paused new data-center tax exemptions during legislative review. Maryland’s consumer advocate challenged regional transmission cost-shifting at FERC. Utah added phased review and environmental scrutiny to a major proposed data-center project.

These are not anti-AI actions. They are governance actions. They suggest that local and state institutions are beginning to insist that AI infrastructure be subject to public rules around land, water, power, cost, disclosure, and community impact.

Public interpretation: AI’s physical footprint is becoming one of the first places where democratic institutions can contest the default concentration path.

2. Colorado showed both governance rollback and local catch-up

Direction: Mixed
Confidence: High

Colorado produced some of the most interesting and contrasting signals of the month.

At the state level, Colorado narrowed its earlier AI governance framework, shifting toward a more limited automated-decision law built around notice, adverse-action procedures, human review, and record retention rather than broader systemic transparency. The event log scores that as Concentration +2 because it reduces public/systemic disclosure and moves accountability toward individualized remedies.

At the same time, Colorado local governments moved in the opposite direction on data centers. Denver and Jefferson County adopted moratoria, and the failed statewide data-center bills may push more governance activity to local land-use channels.

Public interpretation: Colorado is a useful case study in the race itself: state-level AI regulation narrowed, while local compute governance strengthened.

3. AI deployment capacity became a concentration layer

Direction: Concentration-leaning
Confidence: High

OpenAI’s Deployment Company was May’s strongest concentration signal. The event log scores it as Concentration +3, because it moves the frontier AI company beyond model access into workflow transformation, consulting, engineering, and organizational redesign.

That matters because implementation capacity may become as important as model access. If organizations depend on a small number of AI firms not only for models but for redesigning how they work, concentration deepens. The dependency becomes operational, not just technical.

Public interpretation: The next concentration layer may be deployment itself: who has the expertise, staff, tools, and institutional access to make AI usable inside organizations.

4. Public workforce adaptation began to appear

Direction: Dispersion-leaning, early
Confidence: Medium

Utah’s statewide AI Workforce Credential and California’s executive order on AI workforce disruption both point toward public institutions trying to distribute AI capacity and prepare for transition shock.

Utah’s credential matters because public higher education can become a broad AI-literacy and workforce-adaptation channel. California’s order matters because it treats worker disruption, small-business adaptation, early-warning systems, and benefit-sharing as public responsibilities rather than private side effects. The event log treats California’s order as a Dispersion +2 signal but notes that its strength depends on follow-through.

Public interpretation: These are early signs of public transition machinery. They are promising, but not yet proof that workers and smaller institutions will gain real power.

5. AI-linked restructuring moved from theory toward evidence

Direction: Concentration-leaning, early
Confidence: Medium

Groupon’s AI-native restructuring is not yet a major scorecard event, but it is worth tracking. The company’s announced job cuts and AI-native redesign support the emerging “lean firm” hypothesis: firms may try to maintain or expand output with fewer workers by rebuilding workflows around AI.

This is still an early signal. It could prove to be ordinary cost-cutting with AI language attached. But if more firms produce sustained revenue or productivity gains with smaller workforces, this becomes a major transition-shock pattern.

Public interpretation: The labor effects of AI are moving from speculation toward observable restructuring decisions.

What to Watch Next

1. Do moratoria become rules?
   The key question is whether Denver, Reno, Jefferson County, Hill County, Harlingen, Huron County, and similar jurisdictions turn temporary pauses into enforceable standards.
2. Does compute governance expand beyond land use?
   May’s strongest signals involved water, power, grid costs, tax incentives, ratepayer exposure, environmental review, and community burden. Watch whether future policy treats data centers as public-infrastructure questions, not just zoning questions.
3. Does state preemption appear?
   Local governments are beginning to assert authority over data centers. If states or courts limit that authority, local dispersion could become a concentration story.
4. Does California’s workforce order produce real machinery?
   The order matters only if it leads to concrete worker-transition systems, small-business supports, benefit-sharing proposals, public reporting, or legislative action.
5. Does deployment capacity become locked to frontier AI firms?
   OpenAI’s Deployment Company should be watched closely. The question is whether implementation knowledge diffuses broadly or becomes another proprietary layer controlled by a few firms.
6. Do AI-native restructurings spread?
   Groupon is a modest signal. It becomes more important if similar firms explicitly connect AI adoption to lower headcount, workflow redesign, margin expansion, or new labor-light operating models.
7. Does public engagement become real civic infrastructure?
   California’s AI-supported public engagement process is promising. The test is whether participation is broad, the methodology is transparent, and the results influence policy.

What the Scoreboard Shows

The Scoreboard will provide a current reading of the race.

Each update will include:

• Current reading: whether the recent pattern leans toward concentration, dispersion, or a mixed outcome.
• Interpretation: why the pattern looks that way.
• Confidence level: how settled or provisional the reading is.
• Scored signals: the developments feeding into the judgment.
• What to watch next: the questions that may change the reading over time.

Scores are provisional. The scoring methodology is explained separately.

They can change as more information becomes available.

That is the point: democratic judgment should be structured, but it should also be revisable.