Q2 2026 was our strongest quarter yet as FLock.io transitioned from research and early pilots into production-scale deployments with governments, enterprise and global institutions. The company has moved from validating core concepts into actively delivering federated learning and privacy-preserving AI infrastructure directly into enterprise operations, as well as beginning to build sovereign AI for governmental agencies.
During this quarter, we advanced our product capabilities, research output and institutional adoption. Our network expanded its training and inference gateways, established sovereign AI initiatives at a state level, joined core global policy advisory groups and presented peer-reviewed research at tier-one academic conferences.
AI Arena reached 10,801 training submissions and more than 903,000 validation submissions across 2,002 participants, while FLock.io expanded its institutional implementation pipeline across Southeast Asia, Africa, healthcare, and UNDP-supported projects.
Strategic partnerships & global positioning
FLock.io’s strategic activity in Q2 centred on establishing sovereign AI infrastructure, expanding global node footprints and leading public sector deployment initiatives.
A defining milestone was FLock.io becoming the key technical partner in a government initiative to deploy sovereign AI for Sarawak, the largest state in Malaysia. Formalised through three landmark MoUs signed during strategic UK-Sarawak roundtables across Manchester, Cambridge and Oxford, FLock.io is collaborating with the Sarawak Artificial Intelligence Centre (SAIC) and the University of Oxford. The initiative focuses on deploying federated learning and privacy-preserving infrastructure across Sarawak’s priority sectors. Following the signing, Head of Institutional Development Mo Marikar joined UK MP and Trade Envoy to ASEAN Naz Shah in high-level discussions with Sarawak Premier Abang Johari and SAIC CEO Prof. Patrick Then on AI, sustainable energy, and public infrastructure.
This sovereign AI approach was extended internationally through a strategic expansion into Africa. In partnership with BitValue Capital, FLock.io is supporting the deployment of privacy-preserving AI node centres as part of a $200M digital infrastructure initiative across the continent.
FLock.io joined the UNDP Blockchain Advisory Group alongside the Ethereum Foundation, Cardano Foundation, Avalanche Foundation, Sui Foundation and Stellar Development Foundation to advance responsible blockchain and AI applications for Sustainable Development Goals. In Washington, FLock.io joined DCG and industry leaders in policy discussions around the Great American AI Act, which outlined regulatory frameworks for AI systems on decentralised networks.
These high-level engagements directly translated into operational implementations. Cohort 2 of the SDG Blockchain Accelerator concluded with two projects from FLock.io’s track – focused on crop insurance and secure payments – advancing from prototype stages into real-world deployments with UNDP country offices.
Research & technical leadership
FLock.io was spotlighted by the World Economic Forum (WEF)’s MINDS programme for two NHS trusts using its privacy-preserving AI to tackle major diseases. Both trusts use its federated learning platform to train clinical models while maintaining 100% data sovereignty. Moorfields Eye Hospital and UCLH are using FLock.io for eye disease detection and diabetes management respectively.
During the quarter, the research team achieved major academic milestones with two papers accepted into ACL 2026. The research – spanning “GraphSynth” in the ACL 2026 Main Conference and “CircuitSynth” in ACL 2026 Findings – tackled the long-standing synthetic-data trade-off between diversity and factual accuracy. By applying probabilistic factor graphs and circuit synthesis, FLock.io developed an algorithmic approach that generates high-quality synthetic data outperforming original raw datasets for downstream model training while reducing hallucinations.
More research was accepted at ACM ICMR 2026, co-authored by Chief AI Scientist Dr Zehua Cheng and CEO Jiahao Sun. The paper introduced a framework that disentangles identity from other features in latent space, enabling AI systems to reason over visual data without recognising individual subjects.
This research output was complemented by writing on harness engineering, detailing how to build dependable infrastructure for AI agents operating in production environments.
The research team actively shared these insights globally, with Dr Zehua Cheng presenting at Korea University on trustworthy LLMs, while the launch of the What the FLock.io podcast provided dedicated deep dives into federated learning, sovereign AI, and enterprise adoption.
Network growth & ecosystem traction
The growth of the FLock.io network in Q2 demonstrates deepening participation from builders, validators, and ecosystem stakeholders. AI Arena surpassed 10,801 training submissions and over 903,717 validation submissions across 2,002 active participants. To support the rapid expansion of autonomous systems, a dedicated training task was introduced for Hermes Agent, expanding support for agentic AI development across our decentralised node infrastructure.
This developer momentum is paired with strong long-term alignment across protocol economics. The FLock.io Foundation re-locked 10,000,000 FLOCK to mint gmFLOCK, reinforcing long-term capital commitment to the network. Protocol mechanics continue to drive structural deflation, with over 417,000 FLOCK permanently burned to date.
At the ecosystem level, FLock.io opened a new physical footprint for the builder community by establishing its London office at Encode Club. Beyond providing a physical hub, this space supported localised developer engagement through the Encode Vibe Coding Hackathon, weekly co-working sessions, and dedicated workshops.
Product & economic innovation
Product engineering in Q2 focused on broadening model access, expanding distributed infrastructure, and maturing the economic incentives underpinning model usage.
The FLock.io API Platform integrated several of the industry’s latest frontier open and proprietary models, including DeepSeek V4 Pro & Flash, Kimi K2.6, Gemini 3.5 Flash, and Nano Banana 2. This allows developers to access leading models from a single privacy-first gateway, lowering barriers to building multi-model applications.
At the economic layer, FOMO continued to expand its footprint to address pricing inefficiencies in model inference. Centralised cloud providers and gateways operate under fixed price floors dictated by compute hosting costs. FOMO disrupts this structure by introducing Model Tokens on Aerodrome, extending the AI model economy directly on Base. This was accompanied by the launch of FOMO Reward Season 2 for staked Model Tokens. Users and agents calling models via the FLock.io API Platform stake corresponding Model Tokens, directly converting model usage into economic returns for creators. Stakers who lock FLOCK to mint gmFLOCK benefit from a boosted effect, further unifying usage with protocol value accrual.
Alongside model economics, FLock.io advanced its inference infrastructure through a live demonstration of FLock.io Kit. Operating across a distributed cluster of three Mac Minis, the demo illustrated how inference workloads can be distributed across edge hardware. As AI workloads become increasingly decentralised, this inference infrastructure complements FLock.io's federated training stack to enable production-scale execution.
Institutional recognition & industry visibility
FLock.io maintained high visibility across major global policy forums, technical summits, and international executive networks throughout the quarter.
The leadership team presented at flagship global gatherings including the UN Global Dialogue on AI Governance, the AI for Good Global Summit in Geneva, UN Blockchain Talks, the Jeju Forum, Bayes AI Global Forum, Proof of Talk, and Techub x HSK Chain. Across these forums, FLock.io emphasised that trustworthy AI requires privacy-first infrastructure and that sovereign AI relies on controlling training and execution environments.
This global presence was supported by targeted market activations. In Seoul, FLock.io co-hosted AI Transformation 2026: AX Networking Day with RocketPunch, Korea's largest business networking platform, demonstrating enterprise AI deployment to regional industry leaders. Additionally, CEO Jiahao Sun was featured by techUK discussing enterprise AI infrastructure, while Head of EMEA Tiffany Wang contributed to industry discussions on Animoca Brands’ Key Takeaways and Demia’s The Future of Trust.
Outlook – Q3 2026
Looking ahead, Q3 will focus on converting production-ready infrastructure into expanded enterprise footprint and sustained protocol revenue.
A key priority will be continuing to drive deployment of nodes in Sarawak alongside BitValue Capital, establishing active federated networks in the region. The team will focus on expanding UNDP pilot projects into full country-level workflows while scaling participation across FOMO Model Tokens and the FLock.io API Platform. Research and development on Ontology, DigiTwin and forward-deployed agents will be executed.
Forward-looking statements
This report contains forward-looking statements regarding our business, strategy, financial condition, and expected future performance. These statements can typically be identified by words such as “may”, “will”, “expect”, “anticipate”, “aim”, “estimate”, “intend”, “plan”, “believe”, “potential”, “continue”, “is/are likely to,” or other similar expressions.
Such statements reflect our management's current expectations and are based on assumptions and information available to us as of the date of this report. They are subject to known and unknown risks, uncertainties, and other factors – many of which are difficult to predict and beyond our control – that could cause our actual results, performance, or achievements to differ materially from those expressed or implied. Accordingly, investors should not place undue reliance on any forward-looking statement.
Except as required by applicable law, we undertake no obligation to update or revise any forward-looking statement, whether as a result of new information, future events or otherwise. This report is provided to investors for informational purposes only and should be read together with our accompanying financial statements and other information we provide.
