Appendix B · Mechanism Study · Companion to Two Futures

The Automation Cliff.

The mechanism that produces the divergence in the projection is visible in a single ratio: profit per employee. The ratio is already unequal. AI makes it extreme. This page shows the lever, the arithmetic, and the instrument the platform uses to bend the curve.

Figures 2024 10-K filings
Calculator Marginal tier brackets
Framework Tiered Profit-Ratio Tax

The lever: profit per employee

Every corporate story eventually reduces to one ratio. How much profit does the firm generate for every worker it employs? The higher the ratio, the more of the firm's output flows to capital rather than labor. AI does not invent this asymmetry. It drives it to levels no earlier technology could.

Annual profit per employee

US-listed firms · 2024 filings · sorted high to low

The top of the chart: firms whose business model already requires minimal labor per dollar of output. The bottom: firms that still employ many people per dollar of output. AI pushes every industry toward the top.

Apple generates roughly six hundred thousand dollars in profit for every employee on payroll. Walmart generates roughly seven thousand. The ratio is not eighty-five to one because Apple is eighty-five times more productive in any craft sense. It is eighty-five to one because Apple's model requires dramatically less human labor per dollar of output — and because the owners of Apple capture a share of output that Walmart's business structure cannot sustain. The policy question this page takes up: what happens when every large firm's business model shifts toward the Apple end of that axis, which is what AI-driven automation makes possible over the next decade.

The labor arbitrage

The reason automation accelerates is not ideological. It is arithmetic. For routine cognitive and physical tasks, the AI system costs roughly an eighth of the human equivalent.

Human worker annual, fully loaded

  • Salary$60,000
  • Benefits$15,000
  • Overhead$10,000
Total $85,000

AI system equivalent output, annual

  • Compute$5,000
  • Energy$2,000
  • Licensing$3,000
Total $10,000

Cost ratio 1 : 8.5

The firm that refuses to automate loses to the firm that does. This is not a policy choice. It is a competitive fact at the level of accounting. Every projection in this appendix assumes automation accelerates, because it will. The question the platform addresses is not whether the productivity gain arrives. It is whether the gain reaches anyone beyond the shareholders of the firms that capture it first.

The implication is structural. In a market where the firm with lower employee overhead has a permanent cost advantage over the firm with higher employee overhead, every firm faces the same choice: automate, or be priced out by a competitor that will. A firm with a hundred employees competing against a firm with ten — same product, same market — is not acting on principle if it keeps its workforce. It is choosing to lose. Within a decade, every market that can be automated will be staffed at a fraction of its current employment, because the firms that do not automate will not survive the firms that do.

This is the trap no individual firm can opt out of. Returns to capital exceed returns to labor (Piketty's r > g) not because anyone designed the system to do this, but because the math of competitive capitalism rewards aggressive labor substitution at every level. The firm that automates wins; the firm that doesn't loses. Aggregate that across millions of firms and the result is the labor-share decline the country has been measuring for forty years. The system does not need bad actors to produce concentration. It produces concentration as a function of operating as designed.

The Tiered Profit-Ratio Tax — the subject of the next section — is the mechanism that interrupts this dynamic without restricting market function. By weighting the corporate tax rate by profit-per-employee, the policy equalizes the after-tax cost structures of competing firms. The firm with more workers pays a lower effective rate; the firm that has automated aggressively pays a higher one. The competitive race to the bottom on employment is no longer mathematically rewarded. Firms compete on product, service, and innovation rather than on which one has laid off faster.

The mechanism: Tiered Profit-Ratio Tax

The platform taxes profit-per-employee, not raw profit. Firms that generate output by employing many people contribute at the baseline rate. Firms that generate output by replacing people with AI contribute more — proportional to how sharply they substitute labor for capital. Employment is rewarded. Pure extraction is not.

Marginal tier brackets

Applied to profit-per-employee · multiplied across headcount

Tier 1 · Baseline $0 – $100K / employee 21%
Tier 2 $100K – $250K 28%
Tier 3 $250K – $500K 35%
Tier 4 $500K – $1M 42%
Tier 5 $1M – $5M 55%
Tier 6 · Extraction $5M and above 70%

Brackets are marginal, like income-tax brackets. The portion of each employee's generated profit that falls inside a given bracket is taxed at that bracket's rate. A firm at $580K profit-per-employee pays 21% on the first $100K, 28% on the next $150K, 35% on the next $250K, and 42% on the remaining $80K — per employee — summed across headcount.

Tiered Profit-Ratio Tax calculator

Enter any firm's profit and headcount. Compare to the current 21% flat rate.

$39.0B ± $1B steps
67,317 ± 1K steps
Try a company
Profit per employee $580,878
Highest bracket reached Tier 4 · 42%
Current flat 21% $8.21B existing corporate tax
Tiered Profit-Ratio Tax $12.95B 33.1% effective rate
Added to CPS pool +$4.74B additional shared contribution

This firm sits in Tier 4. Hiring an additional 10,000 workers — at the same profit — would lower profit-per-employee and shift more of the tax liability into lower brackets. The structure rewards employment directly.

Unlike a raw-profit tax, the tiered structure does not punish scale. A large employer with modest margins — like Walmart — sits at the baseline rate of twenty-one percent. A small firm with extreme leverage — say, a one-person AI operation generating half a billion dollars in profit — sits in the top bracket at an effective rate above sixty percent. The tax tracks the asymmetry AI drives, not the revenue line, which is why it remains effective as automation reshapes what the word "company" even means.

Without profit sharing: the spiral

The logic of an economy that automates without a profit-sharing mechanism runs in one direction only. It is the scenario the projection study labels Scenario A, expressed as a causal chain.

  1. 01

    AI replaces workers in affected sectors

    Firms adopt automation not out of ideology but out of cost arithmetic. The eight-point-five-to-one cost ratio is decisive at any meaningful scale.

  2. 02

    Displaced workers lose wage income

    Labor-to-capital substitution is immediate. Labor-to-new-sector reallocation takes a decade or more. In the gap, household income contracts.

  3. 03

    Aggregate consumer spending contracts

    Roughly seventy percent of US GDP is consumer spending. As wages fall and unemployment rises, that base shrinks — not marginally, but structurally.

  4. 04

    Firms face shrinking markets and cut further

    Automated firms depend on the broader consumer economy for revenue. As that economy contracts, the firms automate or lay off further to preserve margins, which accelerates the decline.

  5. 05

    The automated economy destroys its own customer base

    The firms with the highest profit-per-employee ratios — the most automated — collapse last, but they collapse. There is no customer economy to sustain them. The cycle closes.

This is not a failure of AI. AI delivers exactly what it promises: extraordinary productivity per unit of human labor. It is a failure of the distribution mechanism that surrounds AI. The productivity is real. The customer base disappears because the wages that sustained it were the same wages the productivity replaced. Without a mechanism that recirculates the gains, the arithmetic ends one way only.

With profit sharing: what changes

Country Profit Sharing changes one thing: where the productivity gains land. That one thing reroutes the chain entirely.

Scenario A

Without CPS

  • AI replaces labor
  • Gains flow to capital owners
  • Consumer spending contracts
  • Markets collapse
  • Economy unwinds
Scenario B

With CPS

  • AI replaces labor
  • Gains feed the CPS pool via Tiered Profit-Ratio Tax
  • Every adult citizen receives the dividend
  • Consumer spending holds or grows
  • Economy remains durable

The Tiered Profit-Ratio Tax feeds the pool in direct proportion to how aggressively capital substitutes for labor. A firm that spreads output across many workers contributes modestly. A firm that replaces workers with AI while maintaining revenue contributes a great deal. The pool grows precisely to the extent that automation concentrates gains elsewhere — and flows back to every adult citizen as a dividend from the productivity that was captured on their behalf. The mechanism is self-balancing. The faster automation advances, the larger the shared return becomes.

The technology is coming either way. The platform is the instrument that determines whether the productivity AI delivers becomes shared prosperity or concentrated wealth. The choice is structural, not rhetorical. It is made in the tax code or not at all.

The instrument, the outcome, the choice.

This page showed the mechanism. Two Futures shows the outcome. The platform is the choice expressed in policy.