Substrate adoption is calculated, in the illustrative scenario modeled on this page, to save the AI industry up to $4.49 Billion per year. LB Cooperative asks for 20% (Tier 1) or 50% (Tier 2 Sanders Fork) of savings actually measured under a signed agreement. The rest stays with the AI company. This page shows what the scenario calculates for each of eight Big AI counterparties, what we would ask, what they would keep, and where the cooperative share would go.
Section 2 - Per-Company Breakdown
Each card shows the projected annual savings from substrate adoption, the Tier 1 and Tier 2 fee structures, and what the AI company retains at each tier. xAI and Microsoft AI savings are stated as ranges per Truth-Always discipline.
Section 3 - Industry Aggregate
Stacked horizontal bars show LB Cooperative share, Public Sovereign Fund (Tier 2 only), and the AI company's retained savings. xAI and Microsoft AI bars use the midpoint of their stated ranges for visualization purposes only - the range is the correct figure.
NOTE: All bars show the Tier 2 50/20/30 split. All bars represent 100% of savings for that company. Absolute bar widths scaled to longest bar (Google DeepMind $1.75B) for the aggregate view above. xAI and Microsoft bars use midpoint for rendering only - stated as ranges on cards. Receipt: /proofs/verify/w12f3c057SCENARIO
Section 4 - Political Track
Legislative approach - requires an act of Congress
Cooperative approach - no legislation needed
Both approaches arrive at a comparable fiscal endpoint for the public: substantial annual capital flow into a public-benefit vehicle. The LB Tier 2 path does not destroy AI company ownership, does not require an act of Congress, and is available now - not after a multi-year legislative fight. Companies that voluntarily adopt Tier 2 are also exempt from the moral and political pressure Sanders' bill creates, because they have already resolved the public interest claim voluntarily.
Section 5 - Bilateral Non-Exclusivity (BP094 Canon)
LB stays free to work with whichever AI companies work well with LB. AI companies compete for innovator pipeline access. No time constraints - only popularity constraints.
LB never signs an exclusive arrangement with any single AI counterparty. All eight companies on this page remain in the innovation pipeline simultaneously. The cooperative's independence is its structural protection.
Companies that work well with LB get more innovator pipeline referrals. The pipeline is merit-weighted, not time-exclusivity-weighted. A better product gets more cooperative members. That is the constraint.
The Universal Innovation Pipeline layers on top of either Tier 1 or Tier 2. Pipeline access is not contingent on tier choice - both tiers receive full pipeline participation rights. Canon: canon_universal_innovation_pipeline_marketplace_popularity_constrained_layered_on_tier_1_or_2_bp094
Section 6 - Methodology (Collapsible)
Per-company methodology breakdown. Every figure derives from the W12 F3 Cost/Savings Proof (receipt: /proofs/verify/w12f3c057) and the AA-FORMAL-2272 cost ladder. Three adoption-rate scenarios shown for each company.
Savings estimate: $400M (stated as an illustrative scenario calculation, not a projection of any single customer's outcome).
| Scenario | Adoption Rate | Est. Annual Savings |
|---|---|---|
| Low (conservative) | 20% | $280M |
| Baseline | 28% | $400M |
| High (optimistic) | 40% | $560M |
Per-token cost reduction: substrate routes repetitive knowledge-retrieval queries to locally-cached cross-linked DAG entries instead of full frontier model inference. Token savings estimated at 60-80% of RAG query tokens per call.
Receipt: /proofs/verify/w12f3c057Savings estimate: $1.25B (stated as an illustrative scenario calculation, not a projection of any single customer's outcome). Largest API volume counterparty with most direct sensitivity to per-query cost reduction.
| Scenario | Adoption Rate | Est. Annual Savings |
|---|---|---|
| Low (conservative) | 25% | $875M |
| Baseline | 35% | $1.25B |
| High (optimistic) | 50% | $1.75B |
Baseline cost: GPT-4o RAG at $0.01565/query (2,200 input + 300 output tokens). Substrate cost: Haiku-grade routing at $0.0000688/call. Ratio: 227x. AA-FORMAL-2272 cost ladder applied to estimated query volume.
Receipt: /proofs/verify/w12f3c057Savings estimate: $1.75B (stated as an illustrative scenario calculation, not a projection of any single customer's outcome). Highest single-counterparty figure, reflecting Vertex AI enterprise scale and Gemini API volume. Gemini Frontier substrate ON benchmark score: 90.7 (BP073 Sound Barrier card).
| Scenario | Adoption Rate | Est. Annual Savings |
|---|---|---|
| Low (conservative) | 20% | $1.2B |
| Baseline | 30% | $1.75B |
| High (optimistic) | 45% | $2.4B |
| Company | Low | Baseline | High | Adoption Rate (baseline) |
|---|---|---|---|---|
| Meta | $630M | $900M | $1.26B | 30% |
| Cohere | $70M | $100M | $140M | 40% |
| Mistral | $63M | $90M | $126M | 45% |
| xAI | $500M | N/A - range | $1.0B | Insufficient data for single rate |
| Microsoft AI | $600M | N/A - range | $1.0B | Insufficient data for single rate |
xAI and Microsoft ranges are not sensitivity bands - they are hard range estimates due to insufficient public API volume data. Do not extrapolate midpoints as confirmed figures. All other rows use the W12 F3 cost ladder applied to estimated query volumes at stated adoption rates.
AA-FORMAL-2272 cost ladder: available internally. R10 Eyewitness benchmark date: 2026-04-21.
Receipt: /proofs/verify/w12f3c057Section 7 - Adoption Status (Truth-Always - Live State as of 2026-06-24)
This page reports current state. When letters dispatch and signatures land, this section updates with the cumulative cooperative dividend earned that year. The $4.49B figure is a verified-methodology illustrative scenario calculation, not a confirmed contract, not an offer, and not a projection of any customer's actual outcome. The above table is the honest current state.
mnemosyne/src/main/substrate_savings_receipt.ts, commit 6c92aceb, extended 45e4e302), which emits a tamper-evident, hash-chained receipt for every query and reports a NET cost-avoided figure across all traffic that reached a routing decision - including cloud escalations - rather than only a lower bound over the queries the substrate fully handled. That mechanism, not the scenario figures above, is the honest bridge from a marketing claim to a contractual measurement.
Section 8 - Receipts and Cross-References
All numbers on this page trace to one of the receipts below. Canon references are the authoritative policy documents this page operationalizes.
PICK A TIER. SIGN THE COOPERATIVE. BE PART OF THE PROOF.
The substrate is built. The proofs are confirmed. Thirty waves, 2251 tests, 24 proofs. Wife Test: DONE. The only thing left is the conversation.