# Gensyn Financial Model Decentralised L1 protocol enabling trustless, verifiable machine learning computation by aggregating idle global compute (data centres, desktops, mobile). - Canonical: https://finamodel.com/startups/gensyn - Excel download: https://finamodel.com/startup-models/gensyn.xlsx - Category: Crypto/Web3 - Model type: SaaS ARR / Valuation - Funding round: Seed - Funding: $6.5M - Founded: 2022 - Geography: Not explicitly stated; team references Cambridge University, UK-based investors likely suggest UK/global. - Customer: B2B ## About the company Gensyn is a decentralised Layer-1 protocol for trustless, verifiable machine-learning computation. It aggregates unused capacity from data centres, desktops, and mobile devices, aiming to let machine-learning workloads be executed across a broad supply network rather than a centrally operated cloud. The company was pre-seed, with a seed raise implied by its 2022 deck and no operating revenue disclosed. This creates a two-sided bootstrapping problem: compute providers must be rewarded for reliable capacity, while developers need confidence that jobs can be verified and completed at useful cost and scale. Start the model with compute suppliers, available capacity, utilisation, jobs processed, and price per unit of computation. Protocol fees arise from completed jobs, while validator rewards and token incentives support network security and supply. Separate token flows from cash operating costs, and flex adoption, compute pricing, utilisation, rewards, and infrastructure expense across scenarios. ## What's included - 5-year monthly revenue build with stage-appropriate growth assumptions - Full P&L, headcount plan, and operating-expense schedule - Cash-flow statement, runway, and burn-rate tracking - Valuation via exit multiple with a DCF cross-check - Returns analysis with MOIC and IRR - Unit economics including CAC, LTV, payback, and cohort retention ## Product & value proposition - Gensyn is an L1 trustless protocol for machine learning computation. - Core innovation: work verification via (a) probabilistic approximate verifiability using gradient metadata, and (b) deterministic graph-based pinpoint challenge protocol - enabling untrusted compute nodes to be used reliably. - Game-theoretic incentive layer (Truebit-style): slashable deposits + reward bounties ensure honest participation. - Key features: tokenised compute market with up-front cost estimation; foundation model warm-starts; ex-ante work estimation (halting-problem mitigation via neural net graph unravelling); functional encryption for private datasets; parallel distributed optimisation. - Performance claim: verification is 1,350% faster vs best competition; MNIST classification model verified in 4.58 min (Gensyn) vs 61.67 min (replication) and 0.003 days vs 80.19 days (Ethereum). - Positioning: high AI compute scale + high cost efficiency (quadrant chart, competitor positions redacted). ## Market - TAM: $181bn opportunity by 2031 for decentralised AI compute market. - Market driver: computational complexity of SOTA AI systems doubling every 3 months. - Supply inefficiency: >60% of general compute is wasted - data centres 61% underutilised, desktop/laptop/tablet 75% underutilised, mobile 25% underutilised. - Demand driver: SOTA model compute (PFLOP-days × memory GB) scaling rapidly from BERT Base (2018) to Switch Transformer 1.6T (2021). ## Revenue model - Token-based: compute buyers pay in Gensyn's native token; suppliers earn tokens for contributing verifiable compute. - Protocol takes a fee on compute transactions (implied by token model; explicit fee rate not stated in deck). - Reference mentions "Token Valuation Model" as a separate document linked from the deck. - Price reference: approximate cost of maxing out deep learning training on Nvidia V100 (or equivalent) per hour cited as market benchmark; exact dollar figure redacted. - Revenue streams implied: (1) protocol transaction fee on compute marketplace; (2) potentially native token appreciation as protocol utility grows. ## Traction & metrics - >150 ML researchers and engineers interviewed (academia + Seed-to-Series-B companies). - 50 early adopters in curated Discord community, keen to train immediately. ## Competition / moat - Centralised compute: AWS and equivalents described as expensive oligopolists with limited supply incentives. - Chip performance asymptotic - transistor density gains plateauing (chart: Zeppelin SoC, Ryzen, Tegra Xavier, Qualcomm 855, TI TDA4VM, Apple A15). - Competitor quadrant positions (AI compute scale vs cost efficiency): redacted. - Moat sources claimed: first-of-kind scalable verification mechanism (1,350% speed advantage); game-theoretic honesty enforcement; ETH1 miner migration as supply source; functional encryption for privacy; censorship-resistant governance. ## Team & funding ask / use of funds - Founders: - Ben: PhD in ML optimisation; ex-founder of an anonymous digital identity startup. - Harry: ex-Head of Data Research at ML risk-pricing startup (Cytora); background in applied econometrics. - Pre-seed investors: Counterview Capital, id4 ventures, Fair Custodian (listed); angels from DeepMind, Draper Esprit, University of Cambridge. ## Recommended financial model - Archetype + why: **Token protocol / usage-based compute marketplace model**. Gensyn is a crypto-native protocol where revenue accrues via transaction fees on a token-denominated compute marketplace. The right model blends (a) a usage-based SaaS P&L for the protocol operator layer and (b) a token valuation / network-value model (Metcalfe-style or transaction-fee NPV). A standard 3-statement is insufficient alone because token float, staking, and slashing mechanics affect "revenue" recognition. A SaaS ARR frame is the closest analogue for the fee-income layer; token economics require a separate tab. - Forecast horizon & granularity: 5 years (Year 1–2 monthly, Year 3–5 annual); model should phase-gate on roadmap: Phase 1 (early adopters), Phase 2 (foundation models), Phase 3 (ecosystem DApps). - Key drivers & assumptions: - Total addressable compute market ($181bn by 2031) → implied CAGR ~30% from current base - Gensyn protocol market share by phase - Compute supply volume (GPU-hours on network) - Protocol take rate (fee % on compute transactions) - Token price (if modelling token treasury revenue) - Verifier / trainer node count ramp - ETH1 miner migration as supply catalyst (Phase 1 tailwind) - Average compute job size / PFLOP-days per customer - Customer count ramp: 50 Discord early adopters as Year 1 seed; Phase 2 growth - Infrastructure / protocol opex (team, cloud nodes, security audits) - Scenarios (Base / Bull / Bear - which variables flex): - Base: moderate AI compute market growth, 0.1% protocol share by Year 3, 7% take rate - Bull: AI compute spend accelerates (doubling every 3 months sustained), protocol becomes dominant standard, 0.5%+ share - Bear: adoption lags (slow ETH miner migration, competition from centralised incumbents), take rate compressed, token illiquidity - Required sheets / outputs: 1. Assumptions & drivers (toggle sheet) 2. Compute marketplace volume (GPU-hours, jobs, $ value transacted) 3. Protocol revenue (take-rate fees) - P&L by phase 4. Token economics tab (supply schedule, staking, slashing reserves, treasury) 5. Headcount & opex 6. Cash / runway (pre-revenue burn through token raise) 7. Market sizing build ($181bn TAM bridge) 8. Scenario toggle (Base / Bull / Bear) 9. Dashboard / outputs summary ## Frequently asked questions ### Is the Gensyn financial model free? Yes. The Gensyn model is a free Excel download with live formulas. ### Can I change the assumptions? Yes. The workbook is editable and its live formulas recalculate when assumptions change.