NATIX Network has recently introduced its 360Data Subnet (Subnet 72) on the Bittensor decentralized AI network, marking a significant step in its mission to revolutionize mapping and autonomous driving through decentralized, real-world data processing.
Utilizing a Solana-based decentralized physical infrastructure network of smart cameras, NATIX integrates high-volume, street-level data collection with scalable machine learning via Bittensor (TAO), as reported exclusively by crypto.news.
The subnet, developed in collaboration with Bittensor infrastructure partner Yuma, utilizes 360° video feeds captured from NATIX-enabled Tesla vehicles and other mobile devices. This data is then transformed into AI models that enhance real-time map creation and vehicle autonomy.
This move aligns with trends observed in the autonomous mobility sector. Major companies like Uber are investing in synthetic simulation and customized data collections to enhance their self-driving technologies. However, these methods are costly and lag behind real-world developments.
To tackle this challenge, NATIX is targeting a global community of over 250,000 drivers who have collectively covered more than 170 million kilometers through smartphone and VX360 device usage.
VX360 and decentralized data training
The VX360 device, developed in partnership with Grab, utilizes Tesla’s existing camera systems to capture 360° imagery without the need for expensive new hardware. The data is processed both in the cloud and on edge smartphones, detecting elements such as traffic lights and signs in real-time. More complex classification tasks are handled off-device.
Through Bittensor’s decentralized framework, NATIX miners are incentivized to train and enhance AI models on the subnet, which are then deployed across the NATIX Edge Network.
“The NATIX 360Data Subnet on Bittensor represents the convergence of Physical AI and decentralized intelligence,” stated NATIX CEO and co-founder Alireza Ghods. “By decentralizing data analysis, we enable continuous improvement of AI models, significantly boosting mapping precision, autonomous vehicle safety, and real-world responsiveness.”
The primary focus of the subnet is roadwork detection, a crucial application for both mapping platforms and autonomous vehicle navigation. In the long run, NATIX aims to broaden its operations to include pothole detection and infrastructure analysis. Ultimately, NATIX aims to offer comprehensive scenario classification to support autonomous vehicle training.
Q&A with NATIX CEO Alireza Ghods
To gain deeper insights into the impact of NATIX’s recent launch, crypto.news engaged in a conversation with Alireza Ghods, CEO and co-founder of NATIX. The discussion delves into the company’s strategy for scaling decentralized data capture, its competitive stance against tech giants, and its plans to establish demand for its token economy.
The full Q&A is presented below:
crypto.news: Tech giants like Waymo, Tesla, and others, as well as Mobileye, have dedicated significant resources to collecting driving data and advancing autonomous technology. What sets NATIX apart in terms of offering data or AI services of comparable quality? While crowdsourced data enables scale, can crowd-collected camera footage truly match the precision of dedicated mapping vehicles?
Alireza Ghods: Data collection for mapping and autonomous driving is incredibly expensive. Large companies that gather this data do not share it, as they leverage it to gain a competitive edge in their products and services. None of these players have amassed data at the scale achievable by NATIX through crowd-sourcing. The largest open-source dataset for autonomous driving is Learning to Drive (L2D), compiled through a collaboration between two projects using identical sensor suites on 60 EVs from driving schools in German cities over three years, resulting in just 5k hours logged. Since the VX360 launch on May 2nd, 2025, we have already accumulated 2k hours of driving data in just 10 days, with only a third of the pre-ordered devices operational in the US.
While NATIX may not match the precision and quality of dedicated mapping vehicles, the extensive network and data collection compensate for this, yielding results on par with data collected by specialized mapping vehicles. Take Grab, for instance. Grab developed its entire mapping solution based on crowd-sourced efforts, resulting in maps more accurate than Google Maps in Southeast Asia. For crowd-sourced efforts to succeed, a large network of drivers/contributors is essential. With over 250K registered drivers, NATIX boasts the largest camera DePIN network globally.
crypto.news: Collecting 360° video from potentially thousands, if not more, cars and phones in real-time presents a significant technical challenge. How does NATIX manage data ingestion and processing on a large scale?
AG: The processing for ScenGen, which is performed for data collected with the VX360, is computationally intensive and cannot be executed at the edge. We conduct this processing at the cloud level. User-shared data is uploaded only when they are connected to their home Wi-Fi network.
Moreover, we utilize our smartphone network, which possesses edge computing power, to detect map attributes like traffic signs and lights in real-time.
crypto.news: Could you outline the scaling strategy? How does NATIX intend to transition from a pilot program with a few hundred Teslas or phones to a deployment involving, for instance, 100k+ devices streaming data, without incurring system failures or exponential cost increases?
AG: The VX360 device is cost-effective. With a $350 device, we access 360° street-level imagery while offering utility to drivers. Our approach involves leveraging commodity hardware. Nearly everyone owns a smartphone, and Tesla vehicles already feature 360° cameras, making it a matter of tapping into existing resources. If we were to develop a dedicated 360° camera, the hardware cost would be 20-100 times more expensive than VX360 and would lack the scalability of a crowd-sourced solution.
For future scaling, we are in discussions with partners interested in operating fleets of VX360s. These partners could be fleet owners or integrators seeking to sponsor device costs in exchange for reward sharing. For instance, we are engaging with a Tesla fleet owner managing over 3,000 Tesla vehicles within their network. We are also developing fleet management functionalities for fleet operators to enhance the product’s utility and extend beyond crypto rewards.
crypto.news: Building on the previous question, do you view these major players as NATIX’s clients, where you will sell them supplementary data, or as rivals you need to surpass? If the latter, where specifically do you believe your decentralized approach provides an advantage over better-funded operations with full control of their fleets?
AG: We count Grab among our clients and are in the process of finalizing agreements with several leading autonomous driving players, both as direct data clients and partners in developing various autonomous driving products.
Every major player represents a potential customer, either presently or in the future. Some significant players have invested millions in amassing their datasets, which will soon become outdated as road conditions evolve continuously. Even those indicating they do not require our data currently will likely need it in the future.
Moreover, the autonomous driving companies that have invested in proprietary datasets account for less than 20% of the market. The remaining companies either have not invested significantly in data collection and will require NATIX’s data, or they rely on their customers’ data (if any) but lack the ability to enhance their models’ quality for other clients. This poses a significant challenge for these companies in developing a comprehensive autonomous driving stack monetizable across all automotive clients.
crypto.news: NATIX is not the only entity aiming to decentralize physical infrastructure. Past projects like Hivemapper garnered attention several years ago, while DIMO (vehicle data) pursued similar crowd-sourced models. Their journey to success has been arduous, marked by limited user adoption, token devaluation, and other challenges. What lessons has NATIX learned or what approach is it adopting to distinguish itself from struggling projects?
AG: Successful models like Geodnet also exist. I believe the winning formula lies in focusing on protocol revenue and implementing a robust mechanism for token value accrual (e.g., buyback and burn). Many projects falter due to a lack of protocol revenue or inadequate tokenomics (insufficient value accrual). In this aspect, NATIX aligns closely with Geodnet and follows a similar strategy.
It is essential to evaluate each business model individually. For instance, Hivemapper’s forward-facing dashcams capture high-quality street-level imagery for mapping, whereas NATIX’s 360° data not only caters to mapping but also supports autonomous driving and physical AI. 360° video data is considered the pinnacle of street-level visual data, valued 5-10 times more than Hivemapper’s data and opening doors to numerous new mapmaking applications (e.g., side camera-based POI detection).
Furthermore, the 360° data collected can be leveraged for physical AI, a broader domain that gained significant attention following NVIDIA CEO Jensen Huang’s announcement at CES 2025. NATIX offers data diversity suitable for training, testing, and validating autonomous driving stacks and physical AI at a fraction of the cost compared to centralized solutions requiring millions in investment by companies like Nvidia and Uber. This data serves as a catalyst for bridging the simulation-to-reality and reality-to-simulation gap in physical AI development.
This perspective underscores our belief that we are addressing a more significant challenge, one for which the industry is prepared to pay today. It is important to note that vehicles operate with a surrounding camera system, necessitating 360° data for training, testing, and validating autonomous driving stacks, a task impossible with front-facing dashcam data.
crypto.news: Beyond serving as a reward mechanism, what role does the $NATIX token play within this ecosystem? Will data buyers or AI developers need to utilize $NATIX or TAO tokens to access StreetVision’s services? If not, what will drive demand for $NATIX?
AG: These are distinct ecosystems with differing functionalities. NATIX focuses on data curation, while the StreetVision Subnet specializes in specific insight extraction and AI model development.
Each ecosystem generates its unique value and features a value accrual mechanism for its token. Those interested in NATIX data contribute to $NATIX token value accrual via Protocol Revenue and the buyback and burn mechanism. Conversely, individuals seeking insights or AI models from the StreetVision Subnet contribute to $dTAO value accrual (and partially to $NATIX as the data utilized for such insights/models originates from NATIX).
Additionally, $NATIX is employed for protocol governance and network security. This is facilitated through our staking platform, allowing users to engage in protocol governance. Validators in the StreetVision Subnet must stake $NATIX as well.
crypto.news: Your strategic collaboration with Yuma enhances credibility but also raises concerns about centralization. DCG has invested in NATIX, and Yuma has played a pivotal role in launching Bittensor subnets. How do you ensure that neither Yuma nor other influential backers dominate Subnet 72’s direction or rewards?
AG: Our subnet is open to any entity wishing to run a validator. We are collaborating with two other major Bittensor validator players who will also operate validator nodes. We have eliminated staking or $dTAO holding requirements for miners to promote an inclusive ecosystem. As you are aware, miner centralization poses a more significant concern than validator centralization, given that network outputs depend on miners.
crypto.news: What specific use cases can you highlight for the StreetVision Subnet? For instance, is Grab currently utilizing NATIX’s data to update their maps in Southeast Asia, or is this collaboration merely an initial step towards future endeavors?
AG: StreetVision’s initial focus will be real-time roadwork detection, a critical aspect for mapping updates and AV reliability. Additional use cases in the pipeline include pothole detection, road sign identification, littering detection, and infrastructure monitoring. Moreover, StreetVision Subnet will eventually incorporate analysis and classification of driving videos into various scenarios, ranging from routine traffic conditions to rare “edge cases,” as scenario classification plays a vital role in scenario generation for autonomous driving and Physical AI.
Presently, Grab is leveraging NATIX’s data to develop pipelines for the US and EU markets, with active commercial engagement. As the VX360 was recently launched on May 2nd, we are in the initial stages of data collection to build sample datasets required by autonomous driving clients. Commercial discussions are ongoing with numerous autonomous driving companies. Collaborations are also underway with leading autonomous driving research labs to introduce cutting-edge products for simulation-to-reality and reality-to-simulation, with announcements scheduled in the upcoming months.