Sunil Venkataram
Back to garden
•
🌰 seedling
Abundant vs Scarce After AI - The Bifurcation of Post-Scarcity
type/noteaieconomicsabundance

Abundant vs Scarce After AI — The Bifurcation of Post-Scarcity

The two lists

From Sahaj Garg’s essay, a practical bifurcation of what AI abundance does and does not touch:

Approaching zero marginal cost (over 3-10 years)

  • Information synthesis, analysis, and generation. Any question has a better answer than any individual human expert, instantly, free or nearly so.
  • Legal document drafting, contract review, regulatory compliance. The routine apparatus of legal work.
  • Software generation for standard applications. Custom software becomes as cheap as asking for it.
  • Translation and transcription. Real-time, across any language pair.
  • Basic medical diagnosis and triage. AI systems that outperform most human diagnosticians, available to everyone.
  • Financial modeling, planning, and routine advice. What currently costs thousands per hour from a human advisor becomes free.
  • Education content and tutoring. Personalized instruction at any level, infinitely patient, 24/7.
  • Design variations and iteration. Producing 100 design options is as cheap as producing one.
  • Customer support and routine communication. Any interaction that follows patterns.
  • Manufactured goods. Once robotics matures, the same dynamic applies to construction, food production, and many physical goods — this is the most speculative item on the list.

The common thread: these are categories where the final output can be produced primarily from software-accessible inputs and evaluated through software-accessible metrics. Wherever that’s true, marginal cost falls fast.

Staying scarce (or becoming more valuable)

  • Physical presence and embodied experience. Being in a specific place, at a specific time, with specific people.
  • Physical scarcity. Land, pristine environments, raw materials, beachfront, view lots, agricultural land in favorable climates.
  • Health and physical excellence. Athletic achievement, bodily capability, somatic experience.
  • Craft and artisanal production. The “human-made” premium when the buyer specifically values that a human made it.
  • Taste, curation, and judgment. Knowing what to ask for, what’s good, what matters.
  • Social proof and reputation. Being known, vouched for, trusted by other humans.
  • Novel lived experience. Travel, adventure, real relationships, first-hand encounters.
  • Spiritual and contemplative goods. Retreat experiences, contemplative practice, the fruits of sustained attention.
  • Community and belonging. Clubs, movements, shared purpose — things you can only have by genuinely participating.
  • Political representation. Someone fighting for your interests, accountable to you, with real stakes in the outcome.
  • Risk-bearing and skin-in-the-game. The human who signs the liability waiver, takes responsibility, bears the consequences.

The common thread: these are goods whose value depends on the presence, judgment, reputation, body, or accountability of a specific human. No amount of information processing substitutes for being the person who was there, the person who is trusted, the person whose body did the thing, or the person whose identity is on the line.

Why the bifurcation matters

The two lists behave differently, and lumping them together misses the most important feature of the post-AI economy. Three specific consequences:

1. The price mechanism breaks on the abundant side. Markets coordinate production and consumption through prices that reflect scarcity. When marginal cost approaches zero, the price signal loses meaning for that category. What does it mean for legal document review to have a “price” when the marginal cost of one more review is effectively nothing? Traditional economic analysis built around equilibrium prices starts producing nonsense answers for the abundant side.

This is not an argument that “economics breaks.” It’s an argument that a specific part of the economic toolkit — the part that depends on meaningful marginal costs — stops applying to the abundant goods. Other parts (property rights for scarce goods, status economics, preference revelation) remain intact and may become more important.

2. Status and wealth migrate to the scarce side. If a billion people can get equivalent legal advice for free, the thing that differentiates wealthy from non-wealthy is no longer access to legal advice. It’s the things that stayed scarce: the specific beachfront, the exclusive community, the body that works well, the reputation that opens doors, the presence at the event. Status signaling follows scarcity, and scarcity migrated.

This is already happening at the margins. Silicon Valley’s status markers are increasingly about physical fitness, access to exclusive communities, and direct-to-founder access — not about possession of information or devices that everyone has.

3. The two sides interact in unexpected ways. Abundant goods on the first list often increase the value of scarce goods on the second list. When AI can generate infinite design options, having taste to choose the right one becomes more valuable, not less. When software is free to produce, knowing which software to build becomes more valuable. When medical diagnosis is cheap, the relationship with a doctor who knows you personally becomes more valuable. The abundant side doesn’t compete with the scarce side; it often creates demand for the scarce side by removing the bottlenecks that used to constrain it.

Where goods move across the line

The lists above are a 2025-2030 snapshot, not a static division. Goods can move across the line over time, and the movement is usually one-way:

  • Information → abundant. Happened first.
  • Software → abundant. Happening now.
  • Routine professional services → abundant. Happening now (law, accounting, radiology, financial planning).
  • Entertainment content → partially abundant. Generation is cheap; curation and “knowing what’s good” stays scarce.
  • Manufactured goods → will become abundant if/when robotics matures. Today this item is speculative; in 10-15 years it may be as settled as “information is abundant.”
  • Personalized coaching and education → partially abundant. The content becomes free; the human relationship stays valuable (if people want it).

Goods rarely migrate in the reverse direction. Once something becomes abundant, it very rarely becomes scarce again. The only examples are things where abundance caused a different kind of scarcity to emerge (e.g., abundant food made low-calorie, high-nutrient food more valuable — a scarcity within the abundant category).

What changes about investment and valuation

The bifurcation has consequences for how to think about which companies and assets hold value:

Losing companies and assets:

  • Companies whose primary value is producing high volumes of standardized cognitive output — law firms, agencies, consultancies, writing services
  • Assets whose value depended on scarce access to information (gated databases, licensing regimes around information)
  • Business models that depend on continuous human-mediated service delivery for routine tasks

Winning companies and assets:

  • Physical assets in geographically scarce locations (waterfront, productive agricultural land, politically stable jurisdictions)
  • Brands that carry reputational weight humans trust (the “I only buy from them” loyalty)
  • Platforms for community, connection, and real-world coordination (the things you can only have through belonging)
  • Companies that help humans express taste, curation, and judgment at scale (selection-as-a-service)
  • Health and longevity services where the user’s own body is the scarce asset

This is not a “long physical, short digital” trade — both categories contain winners and losers. It’s more a “long scarcity premium, short abundance commodity” trade, which cuts across the physical/digital line in ways that defy simple heuristics.

The honest caveats

  • The lists are not exhaustive and not settled. Reasonable people disagree about which side some items belong on. Creative work is a classic edge case — Garg puts it on the abundant side, others put it on the scarce side. The answer is probably “depends on whether the creative work is valued for its novelty vs. for being made by a human the buyer trusts.”
  • The “abundance” side is not actually free. Zero marginal cost doesn’t mean zero total cost. Someone still pays for compute, data, training, infrastructure. The question is whether end users see near-zero prices, not whether total spending in the category drops.
  • Political economy complicates the migration. Professional licensing, regulation, and incumbent defense can slow the migration of goods from scarce to abundant by decades. Law, medicine, and accounting will feel the transition more slowly than pure-software categories because their regulatory surface is larger.
  • Some things on the scarce list could eventually migrate. If robotics matures fully, some forms of physical presence (companionship, caregiving) might become partially substitutable. The scarce list is the 10-20 year frame, not a permanent category.
Backlinks (2)
Connected Notes