Most AI analysis falls into two useless categories. Either it is too abstract — “AI will transform everything” — or too tactical — “here is how to use ChatGPT for customer service.” What is missing is the strategic middle layer: a clear picture of how AI actually changes competitive dynamics and what that means for where your business is vulnerable and where it is protected.
The key insight that most commentary misses: AI is not uniformly intensifying competition everywhere. It is bifurcating the economy into two fundamentally different zones with opposite dynamics. In contestable digital markets — where output is easy to compare and customers can switch easily — AI is commoditizing the baseline and crushing middle-tier businesses. In physical, relationship-heavy markets, AI is lowering overhead without increasing competitive pressure, and margins are actually rising.
Your strategic position depends entirely on which zone you occupy. And that should shape every AI investment you make.
The Bifurcation: Two Economies Diverging
The Digital Contestable Zone
A market is “contestable” when buyers can easily evaluate alternatives and switch providers with minimal friction. The concept originates from economist William Baumol’s 1982 theory of contestable markets, which argued that the threat of entry — not the number of existing firms — is the true driver of competitive behavior. Digital services have become the most contestable markets in history: a company in Algiers can hire a marketing agency in Lagos, a design firm in Lisbon, or a lean AI-native team operating from a basement anywhere on earth.
In these markets, AI is devastating for middling competitors:
- The baseline is commoditized. Work that used to differentiate a firm — strategy decks, campaign plans, code, analysis, copy — can now be produced at dramatically lower cost by anyone with good AI tools. McKinsey estimates that generative AI could reduce the cost of software engineering tasks by 20 to 45 percent, with companies achieving high adoption seeing productivity gains exceeding 100 percent.
- Small teams match large firms. A handful of people armed with AI can produce output that competes with a 50-person agency. According to NVIDIA’s 2026 State of AI report, industries that have fully embraced AI are seeing labor productivity grow 4.8 times faster than the global average.
- Switching is frictionless. Clients can evaluate alternatives easily because the output is digital, comparable, and portable. This is the textbook definition of a contestable market.
- Price pressure is relentless. When production costs collapse, clients notice — and they demand the savings. Deloitte’s 2025 enterprise survey found that 66 percent of organizations are already reporting measurable productivity gains from AI adoption.
The Physical Protected Zone
Meanwhile, in markets dominated by physical delivery, local presence, and relationship dynamics, AI tells a completely different story:
- The service cannot be digitally replicated. A plumber has to show up at your house. A dentist has to examine your mouth. An HVAC technician has to be in the room. Nobel Prize-winning AI researcher Geoffrey Hinton has acknowledged this directly, noting that physical trade workers face far less risk from AI than knowledge workers.
- Switching is expensive. You cannot hire a plumber from another continent. Local reputation, trust, and availability create natural barriers that digital disruption cannot erode.
- AI reduces costs without increasing competition. Scheduling, dispatch, invoicing, customer communications — AI automates the back office effectively. A CNN report in 2025 found that over 70 percent of tradespeople have tried AI tools, with 40 percent actively using them for proposals, invoicing, and troubleshooting. But these tools do not expose firms to new competitors.
- Baumol’s cost disease works in their favor. As AI makes other sectors dramatically more productive, economy-wide wage pressure rises, making services that cannot be automated relatively more expensive. The plumber’s pricing power increases, not because they did anything differently, but because the economic forces around them shifted.
The Barbell Effect
The result is a barbell-shaped economy — a pattern that analysts and economists are now documenting across multiple sectors:
Left end: Hyper-efficient AI-native small teams in contestable markets — lean, fast, technology-maximizing. These teams extract enormous value from AI because they have no legacy overhead to protect.
Right end: Large enterprises with distribution moats, platform stickiness, and brand recognition that AI cannot erode. These firms use AI to optimize existing advantages, not to create new ones.
The crushed middle: Mid-tier firms in contestable markets — too large to be lean, too small to have distribution advantages. Harvard Business Review reported in September 2025 that AI is fundamentally changing the structure of consulting firms, with the traditional pyramid model giving way to a leaner “obelisk” structure. The consulting industry illustrates the pattern clearly: boutique specialists and global giants are thriving, while generalist mid-market firms face shrinking margins and an estimated 25 percent of traditional consulting skills becoming obsolete by mid-2026.
Baumol’s Cost Disease: The Counterintuitive Advantage
What It Is
In 1966, economists William Baumol and William Bowen published “Performing Arts: The Economic Dilemma,” a study that introduced one of the most counterintuitive insights in economics. They observed that as some sectors become more productive, wages rise across the entire economy. Sectors that do not become more productive become relatively more expensive — not because they are doing anything wrong, but because their workers could otherwise go work in the more productive sectors.
The classic example: it still takes the same number of musicians to play a Schubert quartet as it did in the 18th century. But musicians today earn dramatically more because the alternative opportunities in productive sectors have pulled wages up. UNESCO has noted that this dynamic continues to shape the cost structure of healthcare, education, and the arts worldwide.
How AI Activates It
AI is making parts of the economy dramatically more productive. McKinsey’s landmark 2023 study estimated that generative AI alone could add $2.6 trillion to $4.4 trillion annually to the global economy — comparable to the entire GDP of the United Kingdom. The World Economic Forum’s Future of Jobs Report 2025 projects that the share of tasks performed primarily by humans will drop from 47 percent today to 33 percent by 2030.
This productivity surge is pulling wages up everywhere. Services that cannot be made more productive by AI — physical trades, healthcare delivery, personal services — will become relatively more expensive. This is the opposite of what most AI commentary predicts. Instead of these workers being threatened, their pricing power increases because their service cannot be replicated by AI, and economy-wide wage growth raises the floor.
The U.S. Bureau of Labor Statistics confirms the pattern: job openings in skilled trades are projected to grow significantly in coming years, even as entry-level positions for college graduates stagnate. The market is already pricing in the bifurcation.
Who Benefits
- Tradespeople — Plumbers, electricians, HVAC technicians, contractors. Microsoft’s research on AI vulnerability ranked installation, maintenance, and repair roles among the least exposed, with only 4 to 6 percent automation risk.
- Healthcare providers — Dentists, physicians, physical therapists, nurses. While AI assists with diagnostics and documentation, the core patient care work remains fundamentally human.
- Personal service providers — Accountants (the relationship-heavy kind), lawyers (litigation, not document review), financial advisers. The judgment and trust components of these roles resist automation.
- Local service businesses — Moving companies, cleaning services, landscaping, home repair. Geographic constraints create natural protection.
These professionals get AI’s efficiency benefits — automated scheduling, invoicing, communications — without AI’s competitive threat of new entrants and commoditized output.
Why the Standard Disruption Narrative Is Wrong
The Story We Keep Hearing
The standard narrative: lean AI-native startups will eat the giants. Big companies are slow, bureaucratic, burdened by legacy systems. Startups move fast, adopt new tools, build AI-native from day one. The disruptors win. The incumbents fall.
What It Gets Wrong
Giants are not as vulnerable as advertised. Many have moats that AI cannot erode: embedded client relationships, bundled offerings, platform status, brand recognition. These advantages actually become more valuable as the production layer commoditizes, because distribution becomes the scarce resource. The stock market reflects this reality — the “Magnificent Seven” tech companies continue to account for an outsized share of market capitalization precisely because their distribution advantages compound as AI raises the production baseline.
Startups are not as threatening as advertised. The capabilities they sell are getting cheaper by the month. If your value proposition is “we use AI to produce X cheaper and faster,” you are in the commodity business. Every other AI-native startup makes the same claim. Your margins compress as underlying models get cheaper and more accessible.
The real casualties are in between. The 40-person marketing agency. The 15-year-old IT consultancy. The mid-market software development shop. The design firm built on being reliable and professional. According to industry analysis, specialized consultants now command fee premiums of 30 to 40 percent over generalists — a clear market signal that generic competence is losing value while deep expertise retains it. Mid-tier firms are getting squeezed from below by tiny AI-native teams and from above by giants with distribution advantages they cannot match.
Advertisement
The Three Layers of Business Value
Understanding where your business creates value is essential for diagnosing your competitive position:
Layer 1: Tokenizable Cognition
Drafting, analysis, coding, research, planning, generating variations — any thinking that can be captured as text. AI has made the marginal cost of this work collapse. McKinsey found that four types of cognitive tasks — customer operations, marketing and sales, software engineering, and R&D — account for 75 percent of generative AI’s potential economic value. Companies can now produce effectively unlimited quantities of this work at near-zero incremental cost.
If this is what you sell, you are in trouble. The work is becoming commoditized. Competitors produce it at similar quality, often at lower cost.
Layer 2: Judgment and Accountability
Someone has to decide which drafts are good. Someone has to sign off on the analysis. Someone has to own the outcome if the recommendation is wrong. This requires human judgment and humans authorized to be accountable. AI generates options; it cannot own accountability for decisions.
This layer is not getting cheaper. It is becoming more valuable as Layer 1 floods the market with abundant cognitive output that needs to be evaluated and directed. The EY Workforce Report from December 2025 found that AI-driven productivity gains are fueling reinvestment in higher-value human roles, not simply replacing them — a strong signal that judgment and accountability are appreciating assets.
Layer 3: Physical Execution
Installation, repair, face-to-face caregiving, hands-on service delivery. No matter how good AI gets at generating text, it cannot show up at your house and fix your furnace. UC Berkeley robotics researcher Ken Goldberg has been direct about the timeline: humanoid robots replacing physical workers remains, in his words, a myth. This layer is constrained by the physical world in ways that do not yield to software improvements.
This layer is protected and gaining pricing power through Baumol’s cost disease. As cognitive work gets cheaper, physical work becomes relatively more valuable.
Strategic Implications
Know Your Layer
The most important strategic question is: which layer does your firm primarily operate in?
- Primarily Layer 1 → High risk. Your output is being commoditized. You need to either get radically lean or move up to Layer 2. The WEF estimates that 30 percent of entry-level work hours could be automated — and entry-level cognitive work is the purest form of Layer 1.
- Primarily Layer 2 → Strong position. Double down on judgment, taste, accountability, and client relationships. Use AI to enhance Layer 1 output while charging for Layer 2 value. This is where the 30 to 40 percent specialist premium lives.
- Primarily Layer 3 → Protected. Use AI for back-office efficiency. Do not overspend on “AI transformation” — your competitive advantage is physical presence, not digital capability. The 70 percent of tradespeople already experimenting with AI tools have the right idea: use it for administrative tasks, not core service delivery.
Know Your Market’s Contestability
- High contestability (digital, remote, output-comparable) → Competition will intensify. The WEF projects that 41 percent of employers plan workforce reductions as AI automates tasks. Differentiate on Layer 2 value or get lean enough to compete on price.
- Low contestability (physical, local, relationship-heavy) → Competition stays stable. AI is a pure efficiency play — lower costs, same or higher prices. The Bureau of Labor Statistics’ projection of growing trade job openings confirms this zone’s resilience.
The Timeline Is Now
This is not a future scenario. Deloitte’s 2025 survey of over 3,200 enterprise leaders found that worker access to AI tools expanded by 50 percent in a single year, with roughly 60 percent of workers now equipped with sanctioned AI tools. The bifurcation is accelerating. Mid-tier firms in contestable markets that have not yet chosen a direction — lean down or move up — are running out of runway.
Conclusion
AI is not a uniform force that disrupts everything equally. It is a selective force that reshapes specific types of markets based on their contestability and the layer of value their firms primarily deliver. Understanding this bifurcation — and honestly diagnosing where your business sits within it — is the prerequisite for any meaningful AI investment strategy.
The firms that thrive will not be the ones that adopt AI fastest. They will be the ones that understand what AI changes about their competitive position and invest accordingly. In contestable markets, that means racing to either extreme of the barbell. In protected markets, it means capturing efficiency gains without overinvesting in transformation theater.
The economy is splitting. The only question is which side of the divide you are building for.
Frequently Asked Questions
What is a contestable market, and why does it matter for AI strategy?
A contestable market is one where buyers can easily evaluate alternatives and switch providers with minimal friction or cost. The concept comes from economist William Baumol’s 1982 theory, which showed that the threat of new entrants — not the number of existing firms — drives competitive behavior. In the AI era, digital service markets have become hyper-contestable because output is easily comparable, delivery is remote, and switching costs approach zero. If your business operates in a contestable market, AI dramatically increases competitive pressure by lowering the cost of production for everyone, including potential new entrants.
How does Baumol’s cost disease benefit physical service workers in the AI era?
Baumol’s cost disease describes how wages rise across the entire economy when some sectors become more productive, even in sectors where productivity has not changed. As AI makes cognitive work dramatically cheaper and more productive, economy-wide wages rise. Physical service workers — plumbers, electricians, healthcare providers — cannot be made more productive by AI (a plumber still has to physically be at your house), but they benefit from the rising wage floor. Their services become relatively more expensive and more valued, increasing their pricing power even though their work has not changed.
Is the “middle” of the barbell economy permanently crushed, or can mid-tier firms survive?
Mid-tier firms can survive, but not by staying where they are. The barbell economy rewards extremes: hyper-lean AI-native teams on one end and large enterprises with distribution moats on the other. Mid-tier firms must choose a direction. They can move toward the lean end by dramatically reducing headcount and overhead while maximizing AI tools — essentially becoming a small, high-output team. Or they can move toward the other end by building defensible advantages in judgment, accountability, and client relationships that AI cannot replicate. The firms that get crushed are the ones that try to keep operating as mid-tier generalists while the cost of their core deliverables collapses around them.
Sources & Further Reading
- FAQ on Ad Agencies: Consolidation, AI Disruption 2026 — eMarketer
- Predictions 2026: Marketing Agencies — Forrester
- Baumol Cost Disease: Long-Term Implications — UNESCO
- The State of AI Competition in Advanced Economies — Federal Reserve
- Skilled Trades Initiatives Expand — Facilities Dive
- Construction Workforce Shortage — Academy of Craft Training
- Construction Labor Shortages, Wages — Fixr.com
- 2026 HVAC Industry Outlook — HVAC Industry Journal
- AI Paradoxes 2026 — World Economic Forum
- Marketing Agency Statistics 2026 — SHNO
