On September 1, 2026, Elon Musk made a bold prediction at the G20 Innovation Ministerial: by the end of 2027, AI could perform almost all work that is purely digital. He also suggested that software development could reach “Stockfish level” within roughly a year—making it difficult for humans to compete directly.
The timing may prove wrong, and “all digital work” deserves careful qualification. But the direction is hard to ignore. Design, development, testing and experimentation are getting cheaper. An idea that once required weeks of coordination can now become a working prototype in hours.
That raises an obvious question: if everyone can build, do product managers become less important?
I believe the opposite.
AI lowers the barrier to building a feature. It does not lower the barrier to building a good product. When code and options become abundant, the scarce work moves upstream: choosing the right problem, understanding the market, designing the experience, making trade-offs and taking responsibility for the outcome.
AI will not turn everyone into a product manager. But it will require more people to exercise product ownership.
When Execution Gets Cheaper
Technology is not becoming worthless. Architecture, reliability, security, proprietary data and deep engineering can still create durable advantages. But “we know how to build software” is becoming a weaker moat on its own.
A team may be able to build many more things, much faster. It may also become much faster at building the wrong things.
The scarce questions are increasingly: Whose problem should we solve? Which experience is worth creating? What should we reject? What outcome is the business willing to keep funding?
AI can help answer these questions. Human teams still own the consequences.
Product Ownership Is More Than Building Software
Product = user + problem solved + delivery vehicle + value exchange
A product is a repeatable way to solve a meaningful problem and sustain a value exchange. Software may be the vehicle, and AI may provide the capability. The market ultimately judges whether the result is useful, adoptable and viable.
By “product ownership,” I do not mean a Scrum title or sole decision-making authority. I mean accountability for framing the problem, assembling evidence, making trade-offs with a cross-functional team and measuring the outcome.
AI can summarize interviews, analyze competitors, draw prototypes and write code. It can recommend, rank and even execute decisions. But it cannot own the mandate, organizational context or consequences of an investment decision.
Walmart’s generative AI search illustrates the distinction. Customers can describe a life event—such as hosting a party or preparing for a new baby—and receive help across product categories. The technology is generative search. The product insight is redefining “find an item” as “complete a life task.”
The differentiator is not merely what the model can generate. It is knowing where customers are stuck, how trust should be earned and which parts of the experience should remain under human control.
Three Domains of Product Ownership
AI-era product ownership spans three connected domains.
The first is the business of the product: customers, value proposition, pricing, unit economics and growth. Shipping on time is not enough if customers do not adopt, pay for or stay with the product.
The second is digital product and experience design. A real experience spans interfaces, data, rules, workflows and sometimes offline service. Microsoft Copilot, for example, uses a person’s existing permissions as the boundary for data access. Identity, security and governance are therefore part of the product—not merely technical details behind it.
The third is AI systems and workflow orchestration. Product leaders do not need to become foundation-model engineers, but they need to understand how work should be divided among models, tools, deterministic software and people. They must define evaluation, approval and escalation paths, and decide whether latency, reliability and cost support the business model.
Shipping an AI feature does not make its behavior reliable. Evals, human boundaries and feedback loops must be designed as part of the product.
Three Starting Points, Three Different Gaps
Product managers already work across business, users and technology. Their next step is to move beyond features and releases: understand investment and economics, work directly with AI systems and accept greater accountability for outcomes.
Business practitioners understand customers, workflows and where value is created. Their gap is turning experience into a repeatable product: separating common needs from one-off exceptions, defining what the smallest version must prove and testing it against real adoption.
Engineers know how to turn requirements into reliable systems. Their gap is moving beyond “Can we build it?” to ask: Is this problem worth solving? What would make customers change an established habit? How will the product create sustainable value?
None of these people needs to replace every specialist. Product ownership means connecting their expertise, keeping the trade-offs visible and remaining accountable for whether the result works.
AI will keep making construction cheaper.
The difficult—and valuable—work is deciding what deserves to exist, whom it should serve and how it can become a product people continue to choose.
When execution gets cheaper, judgment becomes more valuable.
Which gap is hardest in your role today: business judgment, experience design or AI workflow orchestration?
References
- Elon Musk’s remarks at the G20 Innovation Ministerial
- Walmart Global Tech: Walmart’s Generative AI Search
- Microsoft Learn: Microsoft 365 Copilot Architecture
Further Reading
- Harvard Business Review: To Drive AI Adoption, Build Your Team’s Product Management Skills
- SVPG: Build To Learn FAQ
Learn more at ffde.ai.