THE ANANDA PROPOSAL FOR UNIVERSAL AI ACCESS - Mohan Ananda

Artificial Intelligence as Essential Infrastructure for the Twenty-First Century

A National and Global Policy Framework

EXECUTIVE SUMMARY

Every major era of human advancement has been accelerated by the democratization of a resource that was previously scarce.

Clean water transformed public health.

Electricity transformed physical productivity.

Telecommunications transformed communication.

The Internet transformed access to information.

Artificial intelligence has the potential to transform access to intelligence itself.

AI is rapidly becoming more than a technology product. It is evolving into an essential intellectual infrastructure through which people will learn, work, obtain healthcare information, conduct business, perform research, communicate, innovate, and interact with governments.

I therefore propose a new national and ultimately global principle:

Every person should have reasonable access to safe, capable artificial intelligence, regardless of income, geography, age, education, disability, or social status.

I call this the Ananda Proposal for Universal AI Access.

The proposal does not advocate nationalizing artificial intelligence.

It does not propose turning AI companies into government-owned utilities.

Nor does it suggest that government should select one official AI system.

Instead, it proposes treating access to a basic level of AI capability as an essential service, much as modern societies have progressively expanded access to electricity, telecommunications, education, libraries, and the Internet.

The underlying model should be:

UNIVERSAL ACCESS + PRIVATE COMPETITION + PUBLIC INFRASTRUCTURE + RESPONSIBLE GOVERNANCE

The United States should lead this transformation by establishing Universal Basic AI, supported by competitive private providers, national compute and energy infrastructure, AI literacy, consumer protections, interoperability, and risk-based regulation.

The ultimate objective should be even larger:

By 2035, every human being who wants access to artificial intelligence should have a practical means of obtaining it.

PART I — THE HISTORICAL MOMENT

1. From Electricity to Intelligence

When electricity was first commercialized, it was not immediately regarded as something every household required.

It was expensive.

Infrastructure was limited.

Access was concentrated geographically.

Applications were initially relatively narrow.

Eventually society recognized something fundamental:

Electricity itself was not merely another product.

It was an enabling infrastructure upon which thousands of other products, industries, and improvements in quality of life would depend.

AI appears to be following a comparable trajectory.

The important distinction is that electricity multiplied humanity’s physical capabilities.

Artificial intelligence can multiply humanity’s intellectual capabilities.

An individual equipped with increasingly sophisticated AI may have immediate assistance with research, education, translation, programming, writing, analysis, business creation, healthcare information, financial literacy, creativity, and problem-solving.

That represents an extraordinary expansion of individual capability.

For this reason, AI should increasingly be considered essential intellectual infrastructure.

2. AI IS NOT LITERALLY ELECTRICITY

The analogy should not be taken too far.

Electricity is a largely standardized physical commodity.

Artificial intelligence consists of constantly evolving models, applications, data systems, computing resources, algorithms, interfaces, and services.

Government therefore should not regulate AI exactly as it regulates an electric utility.

Instead, America should establish a new policy category:

ESSENTIAL DIGITAL INTELLIGENCE INFRASTRUCTURE

The objective should be to guarantee reasonable access to AI—not government ownership of AI.

Government establishes the rules of the road.
Private companies innovate.
Consumers choose.
Markets compete.
Society ensures that nobody is completely excluded.

PART II — THE PROBLEM

3. THE COMING INTELLIGENCE DIVIDE

We have spent decades discussing the Digital Divide.

The AI era creates the possibility of something much larger:

THE INTELLIGENCE DIVIDE

Imagine two equally capable students.

One has continuous access to a sophisticated AI tutor capable of teaching mathematics, science, history, languages, writing, and programming.

The other does not.

Imagine two small businesses.

One has AI performing research, marketing, customer support, software development, financial analysis, inventory optimization, and administrative work.

The other does not.

Imagine two physicians.

One can efficiently interrogate enormous bodies of medical literature and patient information with AI assistance.

The other cannot.

Over time, these differences can compound.

The fundamental inequality of the AI era may therefore become not merely the difference between rich and poor.

It could increasingly become:

those who have access to powerful intelligence tools and those who do not.

Public policy should address this possibility before it becomes deeply embedded in society.

PART III — THE CORE PROPOSAL

4. ESTABLISH UNIVERSAL BASIC AI

The United States should establish a national policy of:

UNIVERSAL BASIC AI — UBA

Every American should have access to a baseline level of capable artificial intelligence.

UBA should provide reasonable access to:

Education and tutoring;
General knowledge;
Research assistance;
Language translation;
Career and employment assistance;
Government information;
Basic financial education;
Small-business assistance;
Accessibility services;
Personal productivity;
And general health information subject to appropriate safeguards.

UBA would not mean unlimited free access to the world’s most computationally expensive models.

Government should establish a reasonable minimum level of service.

Private companies could continue offering premium subscriptions, specialized agents, enterprise systems, advanced models, additional computing capacity, and other commercial services.

The analogy is public education and public libraries.

Their existence did not eliminate private schools, universities, publishers, bookstores, or information companies.

They expanded the educated population and ultimately enlarged markets.

Universal AI could do the same.

5. ONE PERSON — ONE AI ASSISTANT

The long-term objective should be:

Every American should have the option of a personal AI assistant.

That assistant should eventually be capable of helping an individual throughout life.

For a child: An AI tutor.
For a student: A research assistant.
For a worker: A productivity partner.
For an entrepreneur: A business assistant.
For a patient: A health-information navigator.
For an elderly person: An accessibility and services assistant.
For a citizen: A government-services navigator.

This should not mean one government-controlled AI assistant.

Citizens should be able to choose among competing certified providers.

6. CREATE AN AI UNIVERSAL SERVICE FUND

America successfully used universal-service concepts to expand telecommunications availability.

I propose an :

AI UNIVERSAL SERVICE FUND — AI-USF

Its purpose would be to ensure that financial circumstances do not prevent Americans from obtaining baseline AI access.

Funding could combine:

Federal appropriations;
Competitive procurement;
State participation;
Public-private partnerships;
Philanthropic participation;
And appropriate contributions from commercial ecosystems benefiting from national AI infrastructure.

The program should be designed carefully to avoid becoming an unlimited entitlement.

Government supports basic access.

The marketplace supplies premium capability.

PART IV — BUILDING THE INFRASTRUCTURE

7. AI REQUIRES PHYSICAL INFRASTRUCTURE

There is sometimes a misconception that AI exists almost magically “in the cloud.”

It does not.

Behind artificial intelligence lies enormous physical infrastructure:

ENERGY → CHIPS → DATA CENTERS → CLOUD → NETWORKS → MODELS → APPLICATIONS → PEOPLE

Every layer matters.

AI infrastructure policy therefore cannot be separated from American energy, semiconductor, telecommunications, cybersecurity, and industrial policy.

Data-center electricity demand is already a major planning issue. The Department of Energy has cited estimates under which U.S. data centers could consume as much as approximately 9% of annual electricity generation by 2030, compared with about 4% in 2023.

America therefore needs an energy strategy capable of supporting the AI era.

8. ESTABLISH NATIONAL AI INFRASTRUCTURE ZONES

The federal government should work with states and private industry to designate:

NATIONAL AI INFRASTRUCTURE ZONES

These would be regions where coordinated development could bring together:

AI data centers;
Advanced electricity generation;
Grid infrastructure;
Energy storage;
Fiber-optic connectivity;
Advanced cooling systems;
Semiconductor infrastructure;
Research universities;
Technical colleges;
And skilled workforces.

Federal permitting for qualifying projects should be accelerated without eliminating necessary environmental, safety, water, or community protections.

AI infrastructure should become a strategic national infrastructure priority.

9. CREATE THE AMERICAN AI COMPUTE GRID

Electricity is available through an interconnected grid.

Computing capacity could increasingly benefit from a comparable concept.

I propose an:

AMERICAN AI COMPUTE GRID

This would not be one enormous government computer.

It would be a federated system connecting qualifying resources across:

National laboratories;
Universities;
Government computing centers;
Commercial cloud providers;
Private data centers;
And future public-private computing facilities.

Compute credits could be competitively allocated to:

Scientists;
Universities;
Medical researchers;
Startups;
Small businesses;
Educators;
Nonprofits;
And strategically important national projects.

America should make computational capability available to innovators who have excellent ideas but do not possess billions of dollars of computing infrastructure.

PART V — EDUCATION

10. AI LITERACY SHOULD BECOME A FUNDAMENTAL SKILL

Universal access without education would be insufficient.

America once made literacy a central educational objective.

Later came computer literacy.

The next requirement is:

AI LITERACY.

Students should learn not simply how to ask AI questions, but how to work intelligently with AI.

They should understand:

Prompting and communication;
Critical reasoning;
Source verification;
AI hallucinations;
Bias;
Privacy;
Cybersecurity;
Appropriate attribution;
Independent thinking;
And the limitations of machine-generated conclusions.

The educational objective should never be:

Let AI think for you.

It should be: Learn to think more effectively with AI.

11. AN AI TUTOR FOR EVERY STUDENT

Every American student should eventually have access to an AI educational assistant.

AI could provide personalized tutoring at a scale impossible through conventional education alone.

A student struggling with algebra could receive unlimited explanations.

A gifted student could move ahead.

A child learning English could obtain personalized language assistance.

A student with a disability could receive specialized support.

A teacher could obtain help preparing lessons and understanding individual student needs.

The teacher should remain central to education.

AI should amplify teachers rather than attempt to replace them.

PART VI — GOVERNMENT

12. CREATE AN AI INTERFACE TO GOVERNMENT

Government itself remains unnecessarily difficult for many citizens to navigate.

People must understand thousands of agencies, forms, websites, programs, and eligibility requirements.

AI can change this.

Every American should eventually be able to ask a trusted public-services AI:

“What Social Security benefits am I eligible for?”

“How do I obtain veterans’ benefits?”

“How do I start a company?”

“What permits do I require?”

“What educational assistance is available?”

“How do I renew this license?”

“What federal programs might help my small business?”

The citizen should not have to understand the organizational chart of government.

Government should organize itself around the needs of the citizen.

AI can become the interface.

PART VII — RIGHTS AND RESPONSIBILITIES

13. AN AI ACCESS BILL OF RIGHTS

Universal AI should be accompanied by basic protections.

Citizens should have rights involving:

Access — reasonable access to basic AI capability.

Choice — no mandatory single government AI provider.

Privacy — appropriate protection of personal information.

Transparency — disclosure when AI is materially involved.

Portability — reasonable ability to move appropriate information between providers.

Security — cybersecurity standards appropriate to risk.

Human Review — human appeal for consequential governmental determinations.

Protection of Children — age-appropriate safeguards.

Accessibility — systems usable by people with disabilities.

Accountability — identifiable responsibility for high-risk AI systems.

14. REGULATE RISK, NOT INNOVATION

AI regulation should be proportional to potential harm.

LEVEL I — LOW RISK

Writing, entertainment, general research, translation and productivity.

Government intervention should be minimal.

LEVEL II — CONSEQUENTIAL

Employment, lending, insurance, education and government-benefit decisions.

Greater transparency, testing, documentation and appeal mechanisms should apply.

LEVEL III — HIGH RISK

Medical diagnosis, autonomous transportation, critical infrastructure, aviation, defense, nuclear systems and comparable applications.

Rigorous validation, cybersecurity, monitoring, fail-safe systems and appropriate human oversight should apply.

The principle should be:

The greater the potential consequence of an AI decision, the greater the required level of accountability.

The United States already has a strong foundation in the NIST AI Risk Management Framework. NIST describes the framework as voluntary, rights-preserving and adaptable across industries, and in 2026 began developing a specific trustworthy-AI profile for critical infrastructure.

PART VIII — GOVERNMENT REORGANIZATION

15. CREATE A NATIONAL AI INFRASTRUCTURE COUNCIL

I do not recommend creating a massive new federal agency.

Instead, the President should establish a:

NATIONAL AI INFRASTRUCTURE COUNCIL

The Council should coordinate relevant activities among:

The White House;
Department of Commerce;
NIST;
Department of Energy;
National Science Foundation;
Department of Education;
Department of Labor;
FCC;
Department of Health and Human Services;
national laboratories;
and appropriate security agencies.

Industry, universities, states and consumer representatives should participate through advisory structures.

16. ADD A FOURTH PILLAR TO AMERICA’S AI STRATEGY

The current U.S. AI Action Plan is organized around three major pillars:

Accelerating AI Innovation
Building American AI Infrastructure
Leading in International AI Diplomacy and Security

I propose adding a fourth:

DEMOCRATIZING AI ACCESS

America should not measure AI leadership only by possessing the world’s most advanced models.

True leadership should also be measured by how broadly those capabilities improve the lives of American citizens.

The objective should therefore become:

BUILD THE BEST AI.
BUILD THE INFRASTRUCTURE.
LEAD THE WORLD.
AND MAKE AI ACCESSIBLE TO EVERY AMERICAN.

PART IX — PROPOSED FEDERAL LEGISLATION

17. THE UNIVERSAL AI ACCESS AND INFRASTRUCTURE ACT

Congress should consider comprehensive legislation establishing the national framework.

The legislation should authorize:

Universal Basic AI.
The AI Universal Service Fund.
The American AI Compute Grid.
National AI Infrastructure Zones.
AI literacy and workforce programs.
AI tutoring initiatives.
Public-service AI infrastructure.
AI interoperability standards.
Privacy and portability requirements.
Risk-based AI governance.
Accelerated infrastructure permitting.
University and research compute access.
Small-business AI assistance.
Rural AI access.
Accessibility programs for seniors and people with disabilities.
International AI development programs.
Cybersecurity and national-security protections.

PART X — A FIVE-YEAR AMERICAN ROADMAP

YEAR ONE — 2027

Establish the Foundation

Enact Universal AI Access legislation.
Create the National AI Infrastructure Council.
Define Universal Basic AI.
Establish national interoperability and security standards.
Launch pilots in schools, libraries, rural communities and government agencies.

Begin AI literacy programs.

YEAR TWO — 2028

Build the Infrastructure

Launch the AI Universal Service Fund.
Begin the American AI Compute Grid.
Designate initial AI Infrastructure Zones.
Accelerate power and transmission development.
Expand university compute access.
Train teachers and federal employees.

YEAR THREE — 2029

Universal Availability

Make baseline AI available to every American who wants it.
Provide AI access through schools and public libraries.
Launch nationwide AI tutoring.
Deploy public-service AI assistants.
Expand programs for seniors, veterans, people with disabilities and disadvantaged communities.

YEAR FOUR — 2030

Integration

Connect AI assistants securely with participating government services.
Expand healthcare-navigation, education, employment and entrepreneurship applications.
Implement interoperability and data portability.
Expand advanced compute access to researchers and entrepreneurs.

YEAR FIVE — 2031

AI-Native America

By this stage America should aspire to a society where:

Every citizen can have an AI assistant.
Every student can have an AI tutor.
Every teacher can have an AI teaching assistant.
Every worker can learn to use AI.
Every entrepreneur can access AI business tools.
Every researcher can compete for advanced computing resources.
Every citizen can navigate government through AI.

PART XI — DEVELOPED COUNTRIES

18. A MODEL FOR OTHER ADVANCED ECONOMIES

Other developed nations should adapt Universal AI Access to their institutions.
Countries with universal healthcare could integrate AI into health-system navigation.
Countries with centralized educational systems could implement national AI tutoring.
European countries could integrate universal access with existing privacy and AI governance frameworks.
Japan, South Korea, Singapore, Canada, Australia, the United Kingdom, European nations and others could establish interoperable national AI-access frameworks.
Ultimately, democratic countries should develop compatible standards allowing trusted AI services to operate internationally.

PART XII — DEVELOPING COUNTRIES

19. DO NOT REQUIRE EVERY COUNTRY TO BUILD FRONTIER AI

The appropriate strategy for developing countries is fundamentally different.

They should not be expected to spend billions of dollars independently training frontier AI models.

The priority should be:

ACCESS, ADAPTATION AND APPLICATION.

The World Bank identifies four foundations for inclusive AI ecosystems:

CONNECTIVITY

Electricity, broadband and digital infrastructure.

COMPUTE

Chips, cloud services and data centers.

CONTEXT

Local data, languages and cultural relevance.

COMPETENCY

Human skills and institutional capacity.

These four foundations provide an excellent framework for developing nations.

20. REGIONAL AI INFRASTRUCTURE

Developing countries could pool infrastructure.

Regional development banks and international institutions could finance:

REGIONAL AI COMPUTE CENTERS

Africa could establish multiple regional hubs.
South Asia could develop shared infrastructure.
Latin America could establish interconnected regional AI facilities.
Smaller nations could purchase secure sovereign computing capacity rather than constructing enormous domestic facilities.
AI development assistance should therefore increasingly combine:

ELECTRICITY + CONNECTIVITY + COMPUTE + LOCAL AI + EDUCATION

This could become one of the major development strategies of the coming decades.

PART XIII — A GLOBAL INITIATIVE

21. UNIVERSAL AI ACCESS BY 2035

The United States should invite other nations, international organizations, technology companies, universities, development banks and philanthropies to establish a:

GLOBAL AI ACCESS INITIATIVE

with one long-term objective:

By 2035, every human being who wants access to artificial intelligence should have a practical means of obtaining it.

This does not mean everyone receives identical computing capacity.

It means geography and poverty should not completely exclude a person from the AI revolution.

22. THE GLOBAL AI DEVELOPMENT FUND

The World Bank, regional development banks, governments, technology companies and philanthropies could create financing mechanisms supporting:

Electricity;
Broadband;
Regional compute;
Cloud credits;
Local-language AI;
AI education;
Teacher training;
Health applications;
Agricultural applications;
And entrepreneurship.

Instead of donating yesterday’s technology to developing nations, the world should help them participate in tomorrow’s technology.

PART XIV — ECONOMICS

23. HOW SHOULD UNIVERSAL AI BE FINANCED?

Universal access should not mean unlimited government spending.

The system should use a mixed model.

Government

Funds basic access, public infrastructure, education, research and underserved populations.

Private Industry

Builds models, data centers, applications and premium services.

Consumers

Pay for premium capabilities beyond Universal Basic AI when they choose.

Businesses

Purchase commercial and enterprise services.

Universities

Receive research grants and compute allocations.

International Institutions

Support developing-country infrastructure.

Philanthropy

Supports educational, humanitarian and underserved-community applications.

This distributes costs while preserving competition.

24. MEASURE THE RETURN, NOT SIMPLY THE COST

Universal AI will require substantial investment.

But infrastructure should be evaluated by its economic return.

The interstate highway system was expensive.

Rural electrification was expensive.

The Internet infrastructure was expensive.

Education is expensive.

Yet each helped create economic activity vastly exceeding the initial expenditure.

If AI materially increases productivity, education, scientific discovery, entrepreneurship, government efficiency and healthcare effectiveness, the economic return from democratizing access could greatly exceed its cost.

PART XV — PROTECTING COMPETITION

25. UNIVERSAL AI MUST NOT CREATE AN AI OLIGOPOLY

One danger of the AI revolution is excessive concentration.
Frontier models require enormous capital.
Compute is expensive.
Advanced semiconductors are scarce.
Data-center construction is increasingly capital-intensive.
Government policy should therefore deliberately preserve competition.
Universal AI contracts should use multiple providers.
Open models should remain part of the ecosystem.
Startups should receive compute opportunities.
Universities should have access to infrastructure.
Consumers should be able to change providers.
Interoperability should reduce lock-in.
Universal access should increase competition rather than entrench incumbents.

PART XVI — HUMANITY AND AI

26. AI SHOULD AUGMENT HUMAN BEINGS

The objective of Universal AI Access is not to replace human intelligence.
It is to augment it.
Calculators did not eliminate mathematics.
Computers did not eliminate human creativity.
The Internet did not eliminate human knowledge.
AI should similarly become a tool through which human beings can become more capable.
The ultimate measure of AI should therefore not be how intelligent machines become.

It should be:

How much more capable human beings become because machines are intelligent.

27. THE MORAL QUESTION

Imagine that in 2035 an extraordinarily capable AI tutor exists.
Should only wealthy children have access?
Imagine AI capable of helping a rural physician understand a difficult medical condition.
Should geography determine access?
Imagine an entrepreneur in Africa, India or Latin America capable of building an important company with AI assistance.

Should the absence of local computing infrastructure prevent that possibility?

Imagine an elderly person who cannot navigate complicated government websites but can simply speak to an AI assistant.

Should that capability be a luxury?

These questions illustrate why Universal AI Access will eventually become more than technology policy.

It will become a question of opportunity.

CONCLUSION

THE NEXT GREAT INFRASTRUCTURE FOR HUMANITY

Human civilization progressed when important capabilities became widely distributed.
Water systems distributed sanitation.
Electrical grids distributed energy.
Telecommunications distributed communication.
The Internet distributed information.

Artificial intelligence can distribute something different:

INTELLECTUAL CAPABILITY.

This does not mean machines become substitutes for human beings.
It means human beings acquire tools capable of dramatically expanding what they can learn, create, understand and accomplish.
The United States has an extraordinary opportunity.
We can build the world’s most advanced artificial intelligence.
We can build the infrastructure required to support it.
We can preserve innovation and free enterprise.
We can establish responsible protections.
And simultaneously we can ensure that the benefits of this extraordinary technology are broadly available.

The guiding principle of the Ananda Proposal for Universal AI Access is therefore simple:

Artificial intelligence should not become a privilege of wealth. It should become an infrastructure of opportunity.

And the ultimate ambition should be equally clear:

AI FOR EVERY PERSON.
AI FOR EVERY STUDENT.
AI FOR EVERY WORKER.
AI FOR EVERY ENTREPRENEUR.
AI FOR EVERY COMMUNITY.
AI FOR EVERY NATION.

UNIVERSAL AI ACCESS BY 2035.

If electricity powered the industrial age and the Internet connected the information age, artificial intelligence may become the infrastructure that empowers the age of human intelligence.

Our responsibility is not merely to create it.

Our responsibility is to make its benefits available to humanity.

DR. MOHAN ANANDA

Dr. Mohan Ananda is an entrepreneur, aerospace scientist, attorney, investor, author, and technology executive. He began his career at NASA’s Jet Propulsion Laboratory working on pioneering space programs and subsequently contributed to work associated with the Global Positioning System at The Aerospace Corporation.

He later became the founding Chairman and CEO of Stamps.com, co-founded Envestnet, and served as Chairman and CEO of NASDAQ-listed Innovative International Acquisition Corp. He currently serves as Executive Chairman of DRAI Health, Inc., an artificial-intelligence healthcare company.

Dr. Ananda holds a B.S. in Mechanical Engineering from Coimbatore Institute of Technology, an M.S. in Aeronautics from Caltech, a Ph.D. in Astrodynamics and Control from UCLA, and a J.D. from the University of West Los Angeles. He is a member of the State Bar of California and is the author of several books addressing entrepreneurship, technology, artificial intelligence, and the immigrant experience.