Last updated: 2026-09-19
Economic modeling and go-to-market research suite for the Gonka Network -- a decentralized AI infrastructure platform with ~98% productive compute and Sprint Consensus (2.3B-parameter Transformer-based PoW).
Generates 5 Excel workbooks (1 master + 4 standalone) that model token emission, price scenarios, fee transition, host profitability, and treasury operations over a 10-year horizon, plus GTM research deliverables (v1.2-v1.4). Built for Gonka leadership, investors, and economic stakeholders.
Network context (as of 2026-07-18): Gonka has raised ~$80M total -- a $12M Bitfury strategic round (Nov 2025, entry ~$0.60/GNK) plus Bitfury's $50M commitment (Dec 2025, the first draw from its $1B ethical-AI fund). In 2026, the GAIB x Gonka partnership opened GNK mining rewards to the open market, with GAIB handling GPU procurement, node operations, and reward routing for H100/H200/B200 hardware. Live network size is far below the ~14,000 H100-equivalent GPUs announced Feb 2, 2026 (the announced peak): as of July 18, 2026, joingonka.ai's live counter shows ~1,178 GPUs active, consistent with tracker.gonka.vip's ~1,214. Both the April 2026 figure of ~4,648 GPUs (~113 participants / ~582 MLNodes) and CoinMarketCap's static "~5,000 H100-equivalents" description are stale -- treat the live counters (~1,200 GPUs) as the active-mining denominator. The previously cited "448+ active hosts" does not match any current source. Real-time network data is available from independent explorers: gonka.gg (free public API), gonkascan.com, gonkahub.com, and tracker.gonka.vip. Workbook adoption assumptions remain Feb-2026 estimates and predate these figures; note in particular that per-GPU mining share and host break-even were computed on the 14,000-GPU denominator -- at a ~1,200-GPU live network, per-GPU share is ~11-12x higher and host break-even GNK price is correspondingly lower (~4x lower than even a 5,000-GPU assumption).
pip install -r requirements.txt
python generate.pyOutput: 5 .xlsx files in output/.
To generate a minimal test workbook (Assumptions tab only):
python generate.py --testDependency: Python 3.x + openpyxl==3.1.5 (sole dependency).
gonka_master_model.xlsx -- Complete model with all 5 economic models, executive dashboard, and full parameter set.
| Tab | Description |
|---|---|
| Documentation | Cover sheet with table of contents and tab descriptions |
| Assumptions | All 70 input parameters, scenario selector, and what-if toggles |
| Emission Schedule | Mining reward decay curve and circulating supply over 10 years |
| Token Price | Multi-scenario price trajectories (Conservative/Base/Aggressive) and market cap |
| Fee Transition | Fee revenue vs. mining emission crossover analysis |
| Host Profitability | GPU host economics, income composition, and breakeven analysis |
| Treasury & POL | Community Pool, protocol-owned liquidity, buyback-burn, and floor defense |
| Dashboard | 8 cross-model KPIs, 3-scenario comparison matrix, and summary charts |
Each standalone contains only the model tabs needed for its analysis, plus dedicated Documentation, Assumptions (filtered to relevant parameters), and Definitions tabs.
| Workbook | Tabs | Use Case |
|---|---|---|
gonka_token_price.xlsx |
5 | Token price scenarios and market cap projections |
gonka_fee_transition.xlsx |
6 | When fee revenue overtakes mining emission |
gonka_host_profitability.xlsx |
7 | GPU host ROI, breakeven price, and churn risk |
gonka_treasury_pol.xlsx |
7 | Community Pool runway, POL returns, buyback-burn impact |
This section is a hands-on guide for non-technical users opening the workbooks in Excel or Google Sheets.
- Open any
.xlsxfile from theoutput/folder. - The Documentation tab (first tab) contains a table of contents with clickable links to every other tab.
- Each model tab has a "Back to Documentation" hyperlink in the top-right corner for easy navigation.
On the Assumptions tab, find the Scenario Selector section:
- Click the Active Scenario cell (blue background).
- Choose from the dropdown: Conservative, Base, or Aggressive.
- The entire workbook recalculates automatically -- all price projections, fee estimates, host income, treasury balances, and dashboard KPIs update to reflect the selected scenario.
Three toggles on the Assumptions tab let you model alternative futures:
| Toggle | Values | Effect |
|---|---|---|
| Buyback-Burn Active | Y / N | Enables or disables the buyback-burn mechanism across all models |
| Deploy POL Active | Y / N | Enables or disables protocol-owned liquidity deployment from the Community Pool |
| Tail Emission Toggle | ON / OFF | When ON, enforces a 10,000 GNK/day emission floor to prevent mining rewards from reaching zero |
Change any toggle and all downstream tabs recalculate instantly.
The Dashboard tab (master workbook only) provides an at-a-glance summary:
8 Key Performance Indicators (Year 10, active scenario):
| KPI | Source |
|---|---|
| Year 10 Circulating Supply | Emission Schedule |
| Year 10 Active GNK Price | Token Price |
| Year 10 Circ. Market Cap | Token Price |
| Year 10 Fee/Emission Ratio | Fee Transition |
| Year 10 Host Net Monthly Income | Host Profitability |
| Year 10 Net Treasury Value | Treasury & POL |
| Cumulative Buyback Burn (% Supply) | Treasury & POL |
| Host Churn Risk Periods (Total) | Host Profitability |
Below the KPIs, a Scenario Comparison Matrix shows 6 metrics side-by-side across Conservative/Base/Aggressive scenarios with color-coded cells (red-yellow-green).
| Cell Color | Meaning |
|---|---|
| Blue background, dark blue text | Input -- editable by the user |
| Black text, no fill | Formula -- calculated automatically |
| Green text | Cross-tab reference -- pulls data from another sheet |
| Red text, light red fill | Warning -- flags danger zones (e.g., churn risk) |
All cells except inputs are locked to prevent accidental formula edits.
- Password:
gonka - To unlock: Review > Unprotect Sheet > enter
gonka. - Google Sheets note: XLSX sheet protection is not preserved when opening in Google Sheets. Formulas remain functional but are unprotected.
All tabs are configured for print:
- Landscape orientation, fit-to-page width
- Margins optimized for A4/Letter
- Headers include tab name; footers include page numbers
- Charts are sized to print cleanly alongside data tables
All models use a shared 32-period time axis: 24 monthly periods (Year 1-2) followed by 8 annual periods (Year 3-10). Each model builds on its predecessors in a dependency chain.
Models the exponential decay of GNK mining rewards from the 680M mining allocation.
- Decay curve, cumulative mined supply, circulating vs. locked supply
- Inflation rate comparison with Ethereum
- 3 charts: emission decay, supply composition, inflation benchmark
Projects GNK price under three macro scenarios using the emission schedule as input.
- Conservative, Base, and Aggressive price bands
- Circulating market cap and fully diluted valuation
- Buyback-burn price floor effect
- 2 charts: scenario price trajectories, price vs. supply dual-axis
Analyzes the crossover point where network fee revenue exceeds mining emission value.
- Fee revenue from compute transactions vs. emission value
- Crossover ratio timeline (ratio > 1.0 = self-sustaining)
- Tail emission impact analysis
- 3 charts: revenue split, crossover ratio, fee vs. emission
Models the economics of running a GPU host node on the Gonka Network.
- Monthly income: mining rewards + fee share
- Operating costs: electricity, hardware depreciation, bandwidth
- Breakeven GNK price and ROI analysis
- Churn risk flagging (periods where hosting is unprofitable)
- 3 charts: income composition, Gonka vs. traditional cloud, breakeven price
Simulates the Community Pool (12% of supply) and protocol-owned liquidity strategy.
- Community Pool inflows, outflows, and runway
- POL deployment returns and impermanent loss
- Buyback-burn cumulative impact on circulating supply
- Floor defense trigger analysis
- 3 charts: treasury composition, depletion timeline, buyback-burn impact
The model is driven by 70 parameters organized into 15 groups, defined in models/parameters.py.
| Group | Count | Examples |
|---|---|---|
| Token Supply | 4 | Total supply (1B), mining allocation (68%), community pool (12%), founders (20%) |
| Emission Parameters | 4 | Epoch length, decay factor, initial reward, halvening schedule |
| Revenue Allocation | 4 | Fee split ratios between hosts, treasury, and burn |
| Price Scenarios | 7 | Conservative/Base/Aggressive price low and high bounds |
| Developer Growth | 4 | Developer adoption curve and compute demand multipliers |
| Fee Transition | 2 | Base fee rate, tail emission rate (10,000 GNK/day) |
| GPU Economics | 4 | Hash rate, power consumption, hardware cost |
| Host Economics | 11 | Electricity cost, bandwidth, maintenance, depreciation |
| POL Parameters | 7 | Deployment rate, target APR, IL assumptions |
| Buyback Parameters | 5 | Buyback rate, burn percentage, price floor trigger |
| Floor Defense | 6 | Defense trigger price, reserve ratio, intervention size |
| Treasury Operations | 2 | Operating budget, grant allocation |
| What-If Toggles | 2 | Buyback-Burn Active, Deploy POL Active |
| veGNK Parameters | 5 | Lock duration, boost multiplier, governance weight |
| Network Parameters | 3 | Node count growth, utilization rate, network capacity |
Each parameter carries a confidence level:
- HIGH -- derived from protocol constants or historical data
- MED -- informed estimates based on comparable networks
- LOW -- speculative projections requiring future validation
All parameters include source citations (whitepaper sections, comparable protocol data, or modeling assumptions) visible in the Assumptions tab.
gonka-tokenomics/
generate.py # CLI entry point (--test flag for quick validation)
requirements.txt # openpyxl==3.1.5
research-model-selection.md # Open-model selection research (v1.2, re-baselined 2026-07-18)
strategy-kimi-k25-agent-inference.md # Kimi agent-inference strategy (v1.2, re-baselined 2026-07-18)
models/
parameters.py # 70 parameters, 15 groups (single source of truth)
generators/
workbook_base.py # Assumptions tab, scenario selector, what-if toggles
styles.py # Named styles, color constants, number formats
chart_utils.py # Chart rendering fix for Excel compatibility
emission.py # Emission Schedule tab (32 rows, 10 cols, 3 charts)
token_price.py # Token Price tab (32 rows, 12 cols, 2 charts)
fee_transition.py # Fee Transition tab (32 rows, 14 cols, 3 charts)
host_profit.py # Host Profitability tab (32 rows, 13 cols, 3 charts)
treasury.py # Treasury & POL tab (32 rows, 14 cols, 3 charts)
dashboard.py # Dashboard tab (8 KPIs, scenario matrix, 3 charts)
documentation.py # Master workbook Documentation tab
cover_sheet.py # Standalone Documentation tab
glossary.py # Standalone Definitions tab (22 terms)
standalone.py # Standalone workbook generator
standalone_config.py # Per-standalone configs, glossary terms, narratives
print_setup.py # Print settings and cell protection
output/ # Generated workbooks (gitignored) + committed GTM docs
gonka_*.md # 11 GTM research deliverables (v1.3-v1.4)
pdf/ # PDF renders of the GTM deliverables
17 charts across 6 model/dashboard tabs. All charts use fix_chart_rendering() to patch the openpyxl application tag for reliable rendering in desktop Excel.
The economic models are grounded in four research documents produced during v1.0, plus five deep-dive analyses:
| Document | Description |
|---|---|
Gonka_Macro_Tokenomics_Research.md |
Macro-level tokenomics analysis: supply dynamics, emission curves, market positioning |
Gonka_Tokenomics_Deep_Analysis.md |
Technical deep dive into all economic mechanisms and their interactions |
Gonka_Tokenomics_Explained.md |
Stakeholder-friendly explainer of the entire token economy |
Gonka_Tokenomics_Fine_Tuning_Recommendations.md |
10 prioritized recommendations for optimizing the token economy (capstone) |
Five specialized research reports in .planning/phases/01-deep-macro-tokenomics-research/research/:
- Protocol-Owned Liquidity -- POL strategies, LP mechanics, IL mitigation
- Real Yield and Buybacks -- Buyback-burn economics, sustainable yield models
- veToken and Governance -- Vote-escrow mechanics, governance incentive alignment
- Fee Transition Stress Test -- Crossover timing under adverse conditions
- GPU Economics and Developer Growth -- Compute market dynamics, adoption curves
whitepaper.pdf-- Gonka Network technical whitepapertokenomics.pdf-- Original tokenomics specification
Milestones v1.2-v1.4 added go-to-market research on top of the economic models:
- v1.2 root docs:
research-model-selection.md(open-model landscape) andstrategy-kimi-k25-agent-inference.md(agent-inference strategy) - v1.3-v1.4 deliverables in
output/: message house, developer personas, competitive feature matrix, provider landscape map, agent pricing analysis, agent-native pitch, objection playbook, channel strategy, partnership playbook, PLG growth model, and the v1.4 engineering backlog (PDF renders inoutput/pdf/)
Freshness note (2026-07-18): the two v1.2 root docs were originally written against Kimi K2.5 (Feb 2026) and have since been re-baselined to the July 2026 model landscape (headers dated 2026-07-18; major claims audit-verified as current). The current landscape they reflect: Moonshot shipped K2.6, then Kimi K2.7-Code (June 2026: 1T MoE, 32B active, 256K context, Modified MIT, coding-agent-focused), and launched Kimi K3 via app/API on July 16, 2026 (2.8T-param MoE, 896 experts with 16 active per token, 1M-token context, native multimodal, always-on thinking mode, $3/$15 per M tokens with $0.30 cached input; full open weights scheduled by July 27, 2026 -- the largest open-weight release to date). K3 debuted #1 in Frontend Code Arena (1,679), ahead of Claude Fable 5 (1,631) and GPT-5.6 Sol (1,618) -- the first open model at the closed-frontier tier. The v1.3-v1.4 GTM deliverables in output/ still predate the ~$80M funding total and the GAIB partnership described above.
Deliverables for the Abu Dhabi (ADI Foundation / Sirius International Holding / IHC) partnership discussion, built on a chain snapshot at epoch 397 (Sept 18, 2026: 494 GPUs, 21 hosts, GNK ~$0.15) and September 2026 market rates.
| File | What it is |
|---|---|
output/gonka_adi_partnership_onepager.md |
One-page pitch: GNK on ADI Chain, an Abu Dhabi inference cluster, "AI on ADI" for developers, verifiable-AI records, 90-day pilot |
output/gonka_uae_fleet_blurbs.md |
Two blurbs (v1): what 20,000 H200 + 10,000 MI300X + 1,000 Cerebras CS-3 earn on Gonka, per GPU and per fleet, with price-impact analysis and demand-filled revenue scenarios |
output/gonka_uae_fleet_blurbs_v2.md |
Same, recalibrated to Gonka founder feedback (MI300X-led, contract-rate baseline, $2 / $3 / $4.50 / $14 price path) |
output/gonka_uae_gov_business_report.md |
Full report: fleet economics, why Gonka fits UAE government programs, business opportunities, deal structure, risks, 12-month plan |
output/uae_fleet_model.py, output/uae_fleet_model_v2.py |
Reproducible estimate models (v1 chain-derived; v2 founder-calibrated). Run with python3 from output/ |
output/datacenter_offer.py |
Datacenter offer harness (below) |
output/research/ |
Fact sheets behind the documents: Gonka network state, accelerator economics, UAE landscape, founder calibration |
PDF renders of all of the above are in output/pdf/.
output/datacenter_offer.py simulates what a datacenter's GPU capacity earns if it joins Gonka, so offers can be drafted from a capacity figure. Datacenters join the live network cumulatively in the order given; each offer sheet shows GNK and USD per GPU per day at several GNK prices, network share, collateral needed to unlock full weight, the GNK price at which mining matches on-demand rental, and the long-term-contract comparison. GPU weights use live medians (H100, H200, B200, A100), founder-implied values (B300, MI300X) and a forward estimate for Cerebras CS-3.
cd output && python3 datacenter_offer.py --dc "Khazna:H200=2000,B300=64" --dc "EHC:MI300X=1000"cd output && python3 datacenter_offer.py --dc "Site:MW=5,type=B300" --gnk 0.15 --gnk 2.0--selftest checks that one H200 on today's network reproduces the founders' 1.3x-on-demand yield.
pdfgen/build_pdfs.py renders every root and output/ markdown file to a styled PDF through pandoc and Typst (title block, running header, page numbers, zebra tables, code and quote styling). Requires pandoc >= 3.1 and typst.
python3 pdfgen/build_pdfs.py- ADI one-pager, UAE fleet blurbs (v1 and founder-calibrated v2), government and business report, reproducible fleet models, datacenter offer harness, research fact sheets
- Typst-based PDF pipeline; all 20 markdown documents re-rendered
- Engineering backlog: 18 must-ship and 14 nice-to-have items (
output/gonka_v14_engineering_backlog.md)
- 10 GTM deliverables in
output/with PDF renders: positioning, personas, pricing, competitive analysis, channels, partnerships, and PLG growth model
- Open-model selection research and Kimi agent-inference strategy (re-baselined to the July 2026 landscape on 2026-07-18; see GTM Research freshness note)
- 5 Excel workbooks: 1 master (8 tabs, 17 charts) + 4 standalone
- 70 parameters with confidence levels and source citations
- Scenario selector (Conservative/Base/Aggressive) with full model chain recalculation
- 3 what-if toggles: Buyback-Burn, Deploy POL, Tail Emission
- Executive dashboard with 8 KPIs and scenario comparison matrix
- Cell protection, print-ready layout, cross-platform validated
- Comprehensive tokenomics research across 4 major documents
- 5 deep-dive analyses on POL, buybacks, governance, fees, and GPU economics
- 10 prioritized fine-tuning recommendations for the Gonka token economy