Robotics x Crypto: Building The Frontier

xmaquina avatar
XMAQUINA
@xmaquinaSep 25, 2025

The Spaces convened founders building at the intersection of robotics and crypto to explore how decentralized infrastructure can accelerate physical AI. James (Peak Network) framed the machine economy stack (identity, wallets, DEX, SDK) for machine-to-machine transactions. Nils (Auki Network) detailed decentralized spatial computing: privacy-preserving visual positioning and performance via millions of venue-scale maps, demonstrated by a live humanoid navigation using phone-shot venue reconstructions, and industry interoperability via Intercognitive. Ali (NATIX Network) explained why multi-camera, crowdsourced street-level data is essential for VPS, training, testing and validation—sharing rapid scale (50–60k hours in months, clients needing >1M hours) and a deflationary buyback-and-burn token model. Mike (GEODNET) covered precise localization (RTK) built by a permissionless node network, with subscriptions driving token burn and dual-hosting synergies (e.g., WingBits). Ammo (Codec Flow) presented open-source operators controlling desktops/robots with cloud simulation and browser-based training. A broader debate contrasted AI’s centralizing forces vs crypto’s decentralization and the need for interoperable standards. Market outlook centered on a near-term “iPhone moment” for robots, China’s hardware/software lead, and investment picks (Unitree, AGIBOT, Figure, Agility). Q&A addressed cross-domain data synergy and global inclusion (Africa), concluding with resources and upcoming launches.

Robotics, Deepin, and Physical AI Roundtable – Summary Notes

Participants and Roles

  • Marissa (Ex Machina co-founder; moderator): Host, structured the AMA, guided topic flow across decentralized spatial computing, token models, and the machine economy.
  • Mercury (Ex Machina co-founder): Added perspectives on decentralization vs centralization, geopolitics, and machine identity/payments.
  • James (Peak Network): L1 for the machine economy; introduced Peak’s infrastructure, Peak ID, SDK, and MachineX DEX; discussed cross-chain abstraction and tokenized coordination.
  • Nils (Auki Network): Building the “real world web” for AI and robots; led discussion on decentralized spatial computing (VPS), privacy, performance, and interoperability.
  • Alireza “Ali” (Natix Network): Crowdsourcing multi‑camera street‑level visual data; use cases across VPS, training/testing/validation for AVs and robots; explained token model (buyback/burn, governance).
  • Mike (GEODNET/Geonet): Precise localization (RTK) infrastructure via Deepin; described token burn model, network scale and customer deployments across drones, agriculture, and autonomous robots.
  • Ammo (Codec Flow): Open-source platform for building, deploying, and monetizing AI “operators” that control desktops and robots; highlighted decentralized data collection, open-source hardware/software, and on‑browser training/fine‑tuning.

Context and Opening

  • The robotics sector is gaining formal recognition (robotics categories listed on CoinGecko/CoinMarketCap; sub-$500M category market cap noted as early-stage). Panel represents a stack of physical AI projects: identity, mapping/localization, spatial computing, vision data, operator execution, and L1 infrastructure.
  • Ex Machina (DAO) offers tokenized exposure to private robotics companies (e.g., Figure, “Electronic,” Agility Robotics) and incubates crypto-robotics projects via Davis Labs; planning a machine economy launchpad and TGE in Q4.

Decentralized Spatial Computing (VPS, Mapping, Privacy, Performance)

  • Nils (Auki) framed spatial computing as teaching devices to understand physical space. Visual Positioning Systems (VPS) compare live camera input against environmental models; by design VPS providers can “see what you see,” raising privacy concerns as AI glasses and in‑home robots proliferate.
    • Privacy: Decentralizing VPS reduces the risk that centralized providers log intimate, persistent home/office visual data.
    • Performance: One global map is impractical; many venue‑specific maps plus protocols for transitioning between them improves scalability and latency (hyper-local compute).
    • Civilization-scale bet: Centralized VPS (e.g., Big Tech visions) are unlikely to win; decentralized, venue‑level maps and interoperable protocols will.
  • Live demos and pipeline: Auki filmed venues (e.g., WOW Summit, Rethink in Hong Kong) with phones, uploaded 200+ recordings to the network, processed in parallel by community nodes. The robot then navigates the site using the phones’ shared spatial understanding. Markers (QR codes) identify the environment and connect devices to the correct venue map.
  • Interoperability: Auki, GEODNET, Peak and others co-founded Intercognitive (intercognitive.com; Executive Chair: Rich Robinson) to coordinate standards so robots seamlessly use GEODNET’s RTK outdoors and Auki’s indoor VPS, with Natix’s camera datasets supporting broad VPS coverage.

Infrastructure and Token Models (Mechanisms, Utility, Deflation)

  • Auki (Nils): Deflationary utility token.
    • Demand side: Access to real-world web is priced in USD terms; users burn an equivalent USD value in tokens (e.g., $1,000 of network usage burns $1,000 worth of tokens).
    • Supply side: A deflationary mint creates fewer new tokens than burned and allocates them to a reward pool. Supply participants stake tokens as reputation for SLA and can be slashed for non-performance; rewards are paid proportionally to contributors (e.g., venues whose maps are used).
  • Natix (Ali): Dual role token.
    • Settlement and governance: Token staking for governance/“deep staking,” voting.
    • Revenue-driven buyback & burn: Protocol revenue (e.g., dataset sales) allocates a large portion to buyback and burn. Fixed supply means increasing deflation as monetization scales. Tokens also reward contributors who provide data.
    • Hardware vs phone data: Natix’s Tesla-connected hardware captures multi-camera surround views (six cameras) for higher-value training/testing/validation data versus single smartphone camera streams; indispensable for multi-camera AV stacks.
    • Scale: Four–five months post-launch, Natix collected 50–60k hours of multi-camera driving data—10x the largest open-source multi-camera dataset (~5k hours, mostly in Germany), with enterprise demand at million-hour scale by 2026.
  • GEODNET (Mike): Burn model and scale.
    • Coverage and permissionless deployment: Prior to GEODNET, professional-grade RTK networks were sparse and expensive ($20–50k per node; $5–10k/yr OPEX). Deepin removes deployment friction—OEMs, dealers, and communities buy nodes, set them up, earn tokens, and sell subscriptions.
    • Pricing and burn: Retail subscription ~$400/year. ~50% to the partner (OEM/dealer), remaining used for buyback and burn; ~80% of purchases allocated to burn. As robot/farm/drone subscribers grow, token supply declines. ~30M tokens burned to date from an initial 1B.
    • ARR: ~$6M ARR; customers include Deep Sand (agri automation), DroneDeploy (mapping), Burro (autonomous carriers), Sensori/Verity (fleet robotic mowing). Centimeter‑level precision enables “turn-on-and-go” operations rather than complex base station setup.
    • Dual-mining hardware: Collaboration with WingBits enables hosts to mine multiple tokens on shared rooftop real estate, maximizing utility and economics across Deepin networks.
  • Peak (James): L1 for the machine economy.
    • Infrastructure and abstraction: Peak facilitates cross-border, chain‑abstracted exchanges of machine-native tokens via MachineX DEX; supports an SDK for robotic integration and Peak ID for machine‑to‑machine identity and direct transactability without intermediaries.
    • Mission: A permissionless OS/computer for the machine economy with 60+ apps across 20 industries and millions of devices; aims at global, borderless digital infrastructure for intelligent machines.
  • Codec Flow (Ammo): Open-source operators platform.
    • Capabilities: Tools/infrastructure for cloud simulation, training, and fine-tuning of models; no-code builders via node-based frameworks; marketplace for operator monetization; fair-launched on Solana; open-source-first approach.

Why Deepin and Crypto Are Essential for Physical AI

  • Development and infrastructure costs for robotics are enormous; centralized firms have struggled to scale beyond limited geographies (e.g., AV programs). Deepin enables:
    • Permissionless coverage buildout, lowering CapEx/Opex barriers across localization, compute, and data collection.
    • Borderless incentivization, tapping global contributors for diverse, edge-case-rich datasets (rain/fog/countries/human behavior like jaywalking or dangerous overtakes).
    • Tokenized, programmatic revenue distribution that sustains networks through market cycles (burn models demonstrated resilience even during crypto pullbacks).
  • Decentralization vs centralization:
    • Mercury and Nils: AI inherently centralizes (data/model training aggregations). When coupled with physical AI, centralized entities wield outsized power (e.g., in theory a single company controlling half of autonomous vehicles becomes a systemic risk). Web3 provides the counterforce—decentralized infrastructure, identity, payments, and governance—to mitigate concentration risks.
    • Peer-to-peer machine commerce: Identity frameworks (e.g., Peak ID) and stablecoins enable robots to transact for services (e.g., a humanoid paying a QR-enabled ride service) with low-latency, real-time settlement.
    • Regulatory and geopolitical context: The West’s slower, costlier infrastructure buildout benefits from Deepin’s model; China’s strong centralized development and strict geospatial regulations limit Deepin’s local role but do not diminish global utility and competitiveness of decentralized approaches elsewhere.

Data Synergy, World Models, and Specialization

  • Cross-domain data synergy: Ali pointed to “world foundational models” that unify synthetic and real-world representations across robotics domains (indoor/outdoor, AV, drones). Projects like Odyssey (ex-Cruise VP) aim to synthesize a world model spanning modalities.
  • On-demand routing: Ammo expects near-term predominance of specialized robots; they will fetch data streams on demand per situation (Auki indoor VPS, GEODNET outdoor RTK, Natix camera data), given current hardware limits. Unifying access layers via interoperable SDKs/APIs becomes critical.
  • Coordination benefits: Nils illustrated the macro-economic upside (e.g., Beijing’s ~2-hour average commute equating weekly GDP losses comparable to building the Great Pyramids) unlocked when vehicles coordinate (potentially obviating red lights) via shared spatial/positional protocols.

Global Outlook: China, US, Europe

  • China’s robotics competence:
    • Nils: Chinese engineers dominate robotics software globally (including in Silicon Valley). Western companies rely substantially on Chinese hardware and data; “You cannot reliably source a single critical robot component in the US” (VC perspective). China delivers both hardware and software (e.g., DJI’s best-in-class stack and APIs) at scale.
    • Mike: DJI example—software and APIs set category standards across consumer and enterprise use cases; China reportedly builds more drones in a day than the US does in a year. Western countries must reshore and leverage Deepin/tokonomics to catch up.
    • Ali: American robotics forecasts often understate China; firsthand exposure (Shanghai) demonstrates more advanced deployment and pace.
  • US: Depth of capital markets and crypto innovation (Deepin) can help accelerate infrastructure and reduce adoption friction.
  • Europe: Perceived slower pace due to bureaucracy; some bright spots in open-source (Hugging Face Robots project). Few standout robotics OEMs compared to US/China.

Investment Views (Anecdotal Opinions)

  • Nils: “AGI, Botton, Unity” (noted as Chinese in discussion). Context suggests he is bullish on Chinese humanoid OEMs.
  • Mercury: Treasury exposure includes “Electronic” (noted massive recent raise and valuation increase), Figure AI (claimed valuation cited at $38B in the discussion), and Agility Robotics; exploring Chinese OEMs and software partnerships (e.g., OS-like providers such as Skill AI/Physical Intelligence as general-purpose stacks for multiple humanoids).
  • Mike: Sees humanoids as exciting; uncertain on investment practicalities for Chinese private companies; believes Ex Machina’s model that unlocks allocations is interesting.
  • Ali: Would choose Unitree and Figure; for accessible exposure he suggests Ex Machina’s token and public equities like Tesla.
  • Ammo: Highlights open-source hardware and software as investable trends (mentions “case scale” doing 3D-printable hardware; emphasizes Hugging Face robots as a strong European software hub). Open-source accelerates AI and robotics similar to its impact on LLMs.

Audience Q&A Highlights

  • Multi-dataset value: Cross-modality and cross-robot data strengthens world foundational models; drones/AVs/indoor robots benefit from shared synthetic/real-world context even if not directly executing each other’s tasks.
  • Specialization vs generalization: Near-term robotics likely specialized; robots will connect to the right data providers on demand (indoor VPS, outdoor RTK, camera streams) via standardized interfaces.
  • Africa inclusion (Nigeria/Lagos and beyond):
    • Marissa: Deepin and blockchain enable global infrastructure funding and revenue distribution regardless of geography. As regulations clarify (stablecoin payment framework; securities definitions), more projects can build and share robot-generated revenues across borders. Deepin can also deliver connectivity and coverage where centralized providers under-invest.
    • Ammo: Fair launches and permissionless funding models open founder opportunities everywhere, letting builders in Africa and other regions bootstrap robotics and data networks via crypto.

Closing and Where to Learn More

  • GEODNET: Follow on X; watch the YouTube channel for customer interviews, deployments, and technical content linking localization to robotics.
  • Auki Network: X handle “Auki Network”; weekly community updates and live demos; Discord (discord.gg link mentioned by Nils) hosts Friday AMAs and demonstrations.
  • Natix Network: natix.network; Discord community; Tesla owners can onboard hardware capturing six-camera trips for high-value datasets.
  • Codec Flow: X “codec open flow”; Discord; open-source collaboration on software infrastructure for operators and robotics.
  • Peak Network: peak.xyz; X @peak; Discord (discord.gg/peaknetwork); weekly AMAs; “Get Real” campaign via portal.peak.xyz; MachineX DEX lists many robotics tokens and enables cross-ecosystem trading.
  • Ex Machina: DAO focused on humanoids and physical AI with tokenized exposure to private robotics; Davis Labs incubates crypto-robotics projects; TGE in Q4; hosting final Genesis auction (pre-TGE access); community call and Discord details on X and xmakena.io.

Key Takeaways and Highlights

  • Decentralized spatial computing is both a privacy safeguard and a performance necessity for robots and AI glasses; venue-level maps and local compute outcompete attempts at centralized “global maps.”
  • Deepin networks (visual data, RTK localization, operator execution, identity/payments) collectively unlock scalable robot deployments by collapsing CapEx/Opex, leveraging global contributors, and sustaining networks via deflationary economics.
  • Token models with real revenue-linked burn (Auki, Natix, GEODNET) demonstrated tangible resilience and alignment—burn increases with usage, rewarding supply-side contributors and constraining supply.
  • Interoperability initiatives (Intercognitive) and L1 machine-economy infrastructure (Peak ID, SDK, DEX) form the backbone for standardized robot data exchange, identity, and payment flows.
  • China’s robotics ecosystem has strong hardware/software execution; the West can leverage Deepin/crypto models to accelerate infrastructure buildout and compete globally.
  • Near-term robotics are specialized; they will dynamically connect to the right datasets (indoor VPS, outdoor RTK, camera feeds) and models; world foundational models are emerging to unify cross-domain simulation and learning.
  • Global inclusion (Africa and other regions) is a core strength of Deepin: permissionless deployment, borderless incentives, and programmable revenue sharing let infrastructure and talent flourish beyond traditional hubs.