More than 6 months after its launch, Swarm has published its first annual letter to stakeholders, offering a detailed look into how the team is approaching decentralized autonomy within the Bittensor ecosystem.
The update positions Swarm as one of the more execution-focused robotics efforts emerging from Bittensor, with a clear emphasis on real-world deployment rather than purely theoretical benchmarks.
The letter, written by Swarm’s leadership, reviews milestones achieved in 2025 and sets out concrete priorities for 2026.
What Swarm is Building
Swarm, which operates on Bittensor’s Subnet 124, focuses on autonomous drone flight control. It’s developing an open-source autonomy layer designed to help drones perceive their environment, plan actions, and execute flight decisions without human intervention.
Unlike closed commercial autopilot systems, Swarm treats flight intelligence as a measurable and competitive asset. Within Subnet 124, validators generate structured flight scenarios, while miners submit flight plans that are evaluated in physics-based simulation. Performance is scored based on speed, energy efficiency, and safety, with higher-performing solutions earning greater rewards.
The long-term objective is to create transparent, auditable drone control software that can be deployed in real operational environments while remaining open to community improvement.
2025 Milestones Highlighted in the Letter: Entry Into the Bittensor Ecosystem
Swarm formally launched Subnet 124 in June 2025, and according to the team, since then has been proving that robotics and physical autonomy problems can be effectively coordinated through Bittensor’s incentive model.
By framing drone control as a competitive task with objective evaluation, the subnet attracted contributors with diverse technical backgrounds and validated the use of decentralized incentives for robotics development.
a. Building the Autonomy Stack
Throughout 2025, the team focused on developing the underlying infrastructure required for real-world autonomy. This included integrating:
1. Flight controller communication,
2. Sensor ingestion and observation pipelines,
3. Onboard inference,
4. Action translation and execution, and
5. Safety monitoring and logging.
The letter emphasizes that this foundational layer is often proprietary in traditional robotics companies. Swarm’s stated intent is to open-source the stack to enable broader experimentation and adoption.
b. Ascend and the Langostino Initiative
Swarm also introduced Ascend, a dedicated robotics branch, alongside Langostino, its first product-oriented effort. Langostino is intended to demonstrate that models trained and selected through decentralized competition can be deployed on real drones under operational constraints.
This initiative represents a transition from simulation-only development toward real hardware validation.
c. Technical Validation Through Incentives
One notable outcome cited in the letter is the production of what Swarm describes as the strongest publicly available LiDAR (Light Detection and Ranging)-based drone navigation model. The result was achieved by incentivizing miners through Bittensor rather than relying on a closed internal research team.
The team presents this as evidence that decentralized coordination, when paired with rigorous evaluation, can produce competitive results in complex robotics domains.
Strategic Priorities for 2026
Having reached this level of maturity, Swarm drew out priorities that emphasizes disciplined execution rather than experimentation in the next phase.
Some of the key priorities includes:
a. Establishing a Physical Robotics Base
A key limitation identified since operations was the lack of continuous physical access to hardware. To address this, Swarm plans to establish a dedicated office and robotics lab in Andorra in 2026.
The goal is to accelerate iteration speed by allowing hardware and software development to co-evolve, particularly during testing and debugging phases.
b. Connecting Simulation to Real-World Deployment
Another focus for 2026 is strengthening the link between simulation performance and real-world outcomes. The team plans to improve how real-world feedback feeds back into evaluation and reward mechanisms, ensuring that progress remains measurable and deployment-driven.
Swarm frames this as a step toward making Bittensor a practical engine for real-world autonomy rather than a purely experimental platform.
c. Early Commercial Validation
The letter outlines plans to pursue an initial reference customer in Andorra. Rather than scaling aggressively, the emphasis is on building a repeatable pilot with clear performance metrics and a defined deployment process.
This approach is intended to establish credibility before expanding to additional regions or customers.
d. Capital Planning
To support its 2026 roadmap, Swarm disclosed plans to raise approximately €400,000 through a mix of bootstrapping and external capital. The funding would support team growth, lab infrastructure, increased testing cadence, and operational runway.
The team reiterated that early commercial revenue would be reinvested into strengthening the subnet and long-term ecosystem sustainability.
Why the Letter Matters
Beyond its technical content, the stakeholder letter stands out for its level of transparency. In a sector where many subnet teams remain opaque, Swarm’s detailed communication provides investors, contributors, and observers with a clearer basis for evaluating progress and intent.
As Bittensor continues to expand beyond purely digital intelligence tasks, projects like Swarm illustrate how decentralized incentives may extend into physical, real-world systems.
Whether this approach can scale remains an open question, but Swarm’s 2025 update suggests the team is prioritizing disciplined execution over rapid expansion as it moves into 2026.
Be the first to comment