The Experts Who Built a Roadmap But Forgot to Add the Road

In September 2025, Nigeria's Federal Ministry of Communications, Innovation and Digital Economy released the updated National Artificial Intelligence Strategy. Eighty pages. Five strategic pillars. Thirty-four strategies. A vision statement about becoming a global leader in ethical and inclusive AI innovation.

Well, technically it isn't.

On closer reading, it is an impressive document about intentions.

This piece is not written to dismiss the effort behind it:

• Over 150 stakeholders contributed.
• The SWOT analysis is honest about Nigeria's infrastructure gaps.
• The risk matrix is more rigorous than most comparable African strategy documents.
• The people who built it clearly understand the landscape they're operating in.

But understanding a landscape and building something that changes it are different activities. And the gap between those two things matters enormously to every builder, researcher, and practitioner who has been waiting for this strategy to tell them something actionable.

What the Document Gets Right

Credit where it's due, the NAIS correctly identifies Nigeria's core AI readiness problems:

• Unreliable power infrastructure.
• Thin data centre capacity.
• Brain drain accelerating faster than talent pipelines can compensate.
• A research and development spend of 0.2 percent of GDP against a global average of 2.2 percent.
• Broadband speeds that rank Nigeria 111th globally on mobile and 140th on fixed broadband.

It also names the governance problem directly, listing governance challenges and regulatory uncertainty as explicit weaknesses in its own SWOT analysis. That kind of institutional self-awareness is not nothing.

The document integrates real legal infrastructure like the National Data Protection Act, the Cybercrimes Act, existing intellectual property law, so we know its governance pillar isn't floating without an anchor. And it even references established international frameworks including the NIST AI Risk Management Framework and the AIGN governance model, rather than reinventing principles from scratch.

For a national strategy document produced by a government ministry, these are genuine strengths.

What the Document Cannot Do

That's where it gets.. problematic. A strategy document that correctly identifies every obstacle and then proposes to address those obstacles through collaboration platforms, awareness campaigns, and centers of excellence has not solved anything. It has simply described the problem in enough detail to appear serious about solving it.

The NAIS has no budget: Not an indicative range, not a percentage of GDP target, not a funding envelope per pillar.

Every strategy — from deploying national high-performance computing centers to launching AI accelerators across Nigeria — is presented without any financial grounding whatsoever. A strategy that says "establish national HPC centers" without stating what that costs or where the money comes from is not a strategy. It is a wish with formatting. It doesn't matter if we spent $100,000 or $10 Billion… we're always within budget.

The thirty-four strategies are presented at equal weight with no sequencing logic. Infrastructure must precede talent development, which must precede adoption, which must precede governance enforcement. That is a real and consequential dependency chain that the strategy doesn't acknowledge.

When everything is a priority, the practical effect is that nothing is.

The KPIs mentioned as a methodology concept in the conclusion are never actually embedded in the strategy pillars. There is no mechanism by which NITDA or NCAIR could be held accountable to a specific outcome that was declared in advance. The accountability is gestural rather than structural.

Schrodinger's Blueprint

The September 2025 document is not the first version. The August 2024 draft was circulated publicly before it, and the August 2024 draft contained a Section 1.5 titled Implementation Plan.

It had four named subsections: Governance Structure and Coordination Mechanisms, Key Milestones and Timelines, Monitoring and Evaluation Framework, and Resource Allocation and Funding Mechanisms.

Every single one of those subsections was blank.

Headers with no content beneath them.

By the document's own account, a four-day minister-led workshop held between April 15th -18th 2024, with over 150 stakeholders produced an implementation roadmap with measurable goals and documented dependencies.

The workshop that produced this strategy was not titled "Developing a National AI Strategy for Nigeria." It was titled "Developing the High-level Strategy and Implementation Plan for a National AI Strategy for Nigeria."

The document states explicitly that the sessions "articulated the national AI vision, measurable impact goals, implementation roadmap, governance structures, and necessary steps to catalyse Nigeria into an AI-driven economy with a clear line of sight on critical dependencies."

That roadmap does not appear in the August 2024 draft.

It does not appear in the September 2025 version.

It has not been published anywhere.

More revealing still, the September 2025 final version kept the introduction paragraph describing the workshop and its implementation outputs, still claiming that those outputs were validated through follow-up engagement sessions between July 2024 and January 2025.

The strategy itself does not contain one.

There is no note explaining the absence.

The document simply moves on.

The August 2024 draft was at least structurally honest, it acknowledged through those blank headers that implementation planning was a required component that hadn't been completed. But it raises a question the document never answers: the workshop that was explicitly convened to produce this implementation plan took place four months before that draft was published. What happened to those outputs?

• The sessions ran for four days.
• Over 150 stakeholders participated.
• The minister attended in person.

By the document's own account the sessions produced measurable impact goals, an implementation roadmap, and clear dependencies. None of that work is visible anywhere in what was published.

The September 2025 version then makes a lazy attempt to hide it by removing the blank headers, allowing a first-time reader to assume the strategy is more complete than it is. No explanation. No acknowledgment.

What was added in exchange?

A reference to the broadband connectivity program — a national initiative that the document's own cited data shows has not delivered its targets.

A slightly tightened vision statement. An updated Oxford Insight ranking reflecting Nigeria's improvement from 103rd to 94th, attributed specifically to the "vision" dimension of the measurement framework.

In other words, Nigeria improved its AI readiness score primarily by having a more credible-looking strategy document.

The Participation Theater

The NAIS describes itself as the product of inclusive, collaborative, bottom-up policy development. The foreword emphasises the active involvement of multiple stakeholders. The conclusion documents a public review process open to individuals and organisations.

These claims were tested.

The Nigerian Ethical AI Framework, an independently developed compliance and governance architecture for AI systems operating in Nigeria, was submitted directly to the ministry directly via their official email and website application forms.

A volunteer application was made to the Nigerian AI collective the ministry explicitly called for.

Follow-up was made with the AI collective when no response came.

"We are happy you are interested in the work of the collective. Someone from the team will reach you soon. Welcome to the Nigeria AI collective."

That was in June of 2025.

Complete radio silence since then.

When a strategy document claims openness to external input and then produces no documented evidence that external input shaped the outputs, the participation language becomes a form of legitimacy signalling.

Regardless of intent, this creates a misleading impression for international funders and partner organisations whose resource commitments are contingent on the kind of inclusive process the language implies.

In Search of the Road

Before drawing conclusions, I went looking for the counterargument.

The September 2025 NAIS references NITDA's Strategic Roadmap and Action Plan 2.0 as an aligned framework.

The glossary uses official language that implies a structured implementation architecture exists somewhere. And frankly, it would be uncharitable and downright sloppy to argue that 150 experts spent four days with a minister and produced nothing that ended up anywhere.

So I went looking.

NITDA's Strategic Roadmap & Action Plan (SRAP 2.0) exists! 🎉

It is a 33-page document covering NITDA's strategic direction from 2024 to 2027 and at first glance, finding it was a huge relief.

Of course the implementation work went somewhere.

Of course a minister-led workshop with over 150 stakeholders produced outputs that a professional agency then operationalised into a delivery roadmap.

That would be the logical sequence. That would be how this is supposed to work.

So I went through it systematically, with four specific questions drawn directly from what the NAIS promised:

1. Does SRAP 2.0 implement the five NAIS pillars?
2. Do its AI targets map to NAIS strategies?
3. Does it fill the four blank sections from the August 2024 draft — governance structure, milestone timelines, monitoring framework, resource allocation?
4. Is NCAIR's role specified with accountable owners and deadlines.

The answer to all four is

SRAP 2.0 never maps to or implements the NAIS five-pillar structure at all

AI appears as one of six co-equal emerging technologies under a single pillar, listed alongside UAVs, IoT, Robotics, Blockchain, and Additive Manufacturing.

There is no standalone AI section, no mapping to the NAIS architecture, and no treatment of the NAIS as a distinct document requiring dedicated implementation support.

The AI-adjacent targets in SRAP 2.0–40 research papers, a 20 percent commercialisation rate, $17.5 million raised — are generic metrics spread across all six emerging technologies. They do not correspond to any NAIS strategic pillar. They read as digital economy boosterism, not NAIS execution.

On the four blank sections: SRAP 2.0 has no AI-specific governance architecture. It has year-end targets but no phased milestones, no quarterly benchmarks, nothing that constitutes a phased implementation timeline.

Its success indicators exist as large-number graphics throughout the document but there is no monitoring mechanism, no named review body, no reporting cadence, no deviation-response protocol. And there are zero budget figures anywhere in the document. No line-item allocation for AI, for NCAIR, or for any of its eight pillars.

NCAIR appears twice. Once in a retrospective of past achievements. Once as a future deliverable: "repurpose NCAIR to create an AI and Cybersecurity Experience and Incubation Centre", with a deadline of 2024. That deadline has passed.

SRAP 2.0 itself contains no verification of completion. No accountable owner is named anywhere in the document. It treats AI as a component of a broader digital economy agenda. The NAIS, to the extent it exists as a policy document, has no visible implementation spine inside it.

The NAIS references SRAP 2.0 as an aligned framework while SRAP 2.0 references the NAIS once — listed as one of seven policies under NITDA's execution responsibility, sitting between the National Blockchain Policy and the National Data Strategy.

A one-sentence acknowledgement on page 8 is the entirety of SRAP 2.0's treatment of Nigeria's National AI Strategy.

• No pillar of the NAIS is mapped to any pillar of SRAP 2.0.
• No NAIS-specific milestone appears anywhere in SRAP 2.0's four-year plan.
• No budget, no named owner, no delivery architecture exists for the strategy NITDA has formally listed as its responsibility to execute.
• Neither document contains the implementation work the April 2024 workshop was convened to produce.

Measurement-Shaped Objects

SRAP 2.0 is not purely aspirational. It has numerical targets, year-end deadlines, and some differentiation between milestones. This makes it more structured than the NAIS, but that does not make it an implementation plan.

The numbers are real… I hope

• 3 million tech talents trained by 2027.
• 70% digital literacy achieved by 2027.
• 40% reduction in successful cyberattacks.
• 444 innovation spaces established nationwide.

These figures appear throughout the document in large bold graphics, giving the pages the visual grammar of accountability. But visual grammar and accountability architecture are different things, and SRAP 2.0 has the former without the latter.

What is completely absent is everything that would make those numbers meaningful as governance commitments rather than aspirational targets.

There are no budget figures anywhere in the document. Not a single naira or dollar is allocated to any pillar, initiative, or target. Without resource allocation, a strategic plan becomes no more than a wish-list.

There are no named accountable owners. Every initiative floats unassigned. "NITDA will…" is the closest the document gets to accountability. No directorate is named. No officer is responsible. No unit is identified as the delivery lead for any specific commitment.

There is no consequence architecture. Nothing in the document specifies what happens if the 2025 PKI target is missed, or if 3MTT delivers 900,000 talents instead of 3 million. There is no remediation protocol, no escalation path, no accountability trigger. The targets exist in a consequence-free environment.

There is no monitoring mechanism. No review body is established. No reporting cadence is defined. No named oversight structure exists. No third-party audit provision is made. The success indicators appear as graphic design elements throughout the document. They are not connected to any system that would track, verify, or respond to their achievement or failure. Most targets jump directly from the present to 2027 with no intermediate checkpoints. There are no quarterly or semi-annual milestones that would allow course correction before a target is missed entirely.

On the Retrospective

The document's treatment of SRAP 1.0 deserves specific attention because it demonstrates the same pattern operating in reverse — using the appearance of accountability to avoid its substance.

SRAP 1.0 completed 41 out of 64 initiatives. The document presents this as evidence of effective implementation. The 36 percent failure rate is reframed as follows: "the remaining 23 initiatives have been reflected in SRAP 2.0." Incomplete work has been silently rolled into the next cycle without any analysis of why those 23 initiatives failed, what they cost, or what their failure meant for the Nigerians who depended on them.

That is major red flag dressed as continuity planning. In any serious public administration audit, unexplained initiative failure rolled forward without accountability analysis would be a concerning discovery.

The document's treatment of startup funding requires equal precision. NITDA directly assisted 345 startups, secured $400,000 and ₦382 million in grants, and participated in international forums including GITEX and LEAP. These are the documented direct outcomes of NITDA's intervention.

The document then states that "these efforts significantly contributed to the growth of Nigeria's technology innovation ecosystem, aiding Nigerian startups in creating over 10,000 jobs and raising over $5 billion between 2019 and 2023."

Read that sentence structure carefully. NITDA secured $400,000. The ecosystem raised $5 billion. The phrase "significantly contributed to" bridges those two figures without establishing any causal mechanism.

The $5 billion was raised by an ecosystem that NITDA operates within. The implied contribution claim is technically deniable because the document never says NITDA raised $5 billion but functionally it places a $5 billion outcome in the same paragraph as NITDA's $400,000 intervention and connects them with language of causation.

Instead of accurate attribution, we have a rhetorical structure designed to survive challenge while implying credit it cannot substantiate.

The Honest Verdict

SRAP 2.0 would not survive a serious public administration audit because it cannot answer three basic questions that any implementation plan must answer:

• who is responsible
• what resources are committed
• what happens when targets are missed.

The numerical targets give it the appearance of accountability architecture. The absence of budget figures, named owners, consequence protocols, and monitoring mechanisms means the architecture is decorative. Nigeria now has a national AI strategy without an implementation plan and an implementing agency roadmap without implementation infrastructure.

The experts built a roadmap but the road is not in the strategy, it is not in the roadmap.

It has not been found anywhere in Nigeria's published digital governance architecture.

So we have an entire industry flying blind.

What an Implementable Framework Actually Looks Like

The absence of consequence architecture in the NAIS is not a Nigerian peculiarity. It is the characteristic failure mode of national AI strategies globally, and it is more acute in weak institutional environments where the gap between policy declaration and enforcement is structurally wide.

An implementable AI governance framework requires at minimum: a compliance lifecycle with defined entry points, risk classification, third-party audit requirements, and certification outcomes that are binary rather than aspirational.

It requires penalty structures with specific figures attached to specific violations. It requires an enforcement body whose independence is architecturally guaranteed rather than rhetorically asserted — meaning its budget, appointment process, and publication requirements are structurally insulated from the ministries it oversees.

It requires, above all, a consequence architecture.

A system where outcomes trigger responses automatically rather than waiting for a human decision-maker with both authority and political will to act simultaneously.

That combination of authority and will, is rare in any institutional environment. Designing accountability systems that depend on it is designing systems that will fail at the exact moment they are needed.

The NAIS does not have this. It names bodies and aspirations but never describes the bureaucratic procedure by which a non-compliant AI system would be identified, audited, sanctioned, and corrected. It lists an AI Governance Regulatory Body as an outcome without specifying who appoints it, how its budget is protected, or what legal authority triggers its enforcement actions.

This is not a gap that subsequent policy documents will easily fill, because the incentive structure that produced a strategy document without consequence architecture is the same incentive structure that will produce subsequent documents without it.

What This Means for You

If you are a Nigerian AI researcher, builder, startup founder, or practitioner who has been waiting for the NAIS to provide direction, the honest message is this: the direction it provides is limited, and the limitations are structural rather than accidental.

The strategy tells you the government has formally committed to AI development as a national priority. That commitment changes how international funders engage with Nigerian AI projects, it creates institutional hooks that can be used to advocate for specific resources, and it signals to foreign partners that there is a policy environment worth engaging with.

What the strategy cannot tell you is how an AI system you build will be governed, what compliance requirements will actually be enforced, what the real cost of non-compliance will be, or whether the infrastructure commitments in the document will materialise on any timeline relevant to decisions you are making now.

You are, in the most practical sense, operating without a reliable map.

That is not a reason to stop building. It's a reason to build with clear eyes about the environment you are in. The builders who will navigate this period most effectively are not the ones waiting for upstream clarity that may not arrive in actionable form. They are the ones who have done the work to understand the environment's real constraints, its actual enforcement gaps, its genuine opportunities, and are building accordingly.

The NAIS is a useful document, but not a sufficient one. The difference between those two things is consequential, and the people who understand that difference are better positioned than those who don't.

The frameworks that Nigeria's AI ecosystem actually needs — operational compliance architecture, automated consequence systems, governance structures that survive political cycles — are not waiting for government to produce them. They can be developed independently, tested in practice, and pushed into the policy conversation from below rather than received from above.

That work is already underway. The question is whether enough people in this ecosystem understand clearly enough what they're building under to demand something better, and to build it themselves when the demand goes unanswered.

The Nigerian Ethical AI Framework (NEAIF) and Governing by Data are independently developed frameworks addressing the compliance and accountability gaps identified in this analysis.

Both are publicly available.