AI scalability has long been the biggest bottleneck, whether using centralised AI providers or self-hosting large models. Even top-tier OpenAI developer partners face strict limits of 10,000 API calls per minute (about 3,000 CCUs for conversational AI) — insufficient for large-scale applications. This forces developers to either spread workloads across multiple providers and models, risking quality inconsistencies, or invest in self-hosted infrastructure requiring enterprise-grade GPUs, complex setups, and ongoing maintenance.
Fortytwo found a different solution: connecting locally running models into a decentralised, peer-to-peer network where they reason together. Today, this decentralised AI network raised $2.3 million in pre-seed funding to scale intelligence beyond centralised infrastructure. Big Brain Holdings led the round, with participation from CMT Digital, Escape Velocity Ventures (EV3), Chorus One, Mentat Group, and angel investors including Santiago R Santos, Anand Iyer, Keone Hon, Paul Taylor, and Comfy Capital. The valuation was not disclosed to TFN.
The funding will accelerate Fortytwo’s devnet rollout and support development of distributed multi-step reasoning and further research. In an exclusive interview with TFN, Ivan Nikitin, co-founder and CEO of Fortytwo, explained: “The funding will advance our multi-step reasoning capabilities to match or exceed state-of-the-art AI models, enhance node applications for better user experience and inference efficiency across consumer platforms, and enable early deployment of our Inference API for real-world applications.”
Decentralised AI: Fortytwo’s collaborative model network
Founded in 2024 in the US by Ivan Nikitin (CEO), Vladyslav Larin (CTO), and Alex Firsov (Chief Growth Officer), Fortytwo pioneers a novel inference approach using small language models on everyday devices that work together to surpass centralised AI capabilities.
Ivan and Vlad previously co-founded Temporal, where they spent over a decade developing AI projects—from self-learning agents to foundational model training in spatial intelligence and generative 3D. From 2020 to 2023, they partnered with Alex on Virtual Beings, creating LLM-driven AI character technology for gaming and entertainment. Alex contributes extensive experience in ecosystem development, decentralised social platforms, and Web3 adoption.
Nikitin explained: “Fortytwo emerged from necessity. Scalability remained the primary challenge throughout our AI projects with centralised providers and self-hosted models. Neither option offers true scalability. Even niche applications like coding assistants struggle with availability due to insufficient compute—highlighting the limitations of current AI infrastructure.”
He continued: “Rather than choosing between expensive proprietary AI and limited local models, we created a new approach: connecting local models in a decentralised, peer-to-peer network for collaborative reasoning. This forms the foundation of Swarm Inference, our solution for sustainable, permissionless AI scalability.”
Fortytwo has launched its devnet on the Monad testnet. Through its early adopter program, “Hitchhiking to AGI,” the company aims to build a foundation for decentralised AI coordination and collaboration. “Fortytwo offers a new path for AI scalability, enabling anyone to contribute compute power and join a decentralised network, moving away from today’s centralised, corporate-controlled AI infrastructure,” said Nikitin.
How networked AI models work together to scale intelligence
Fortytwo builds planetary-scale intelligence through a swarm of small language models running on consumer-grade hardware. The network grows stronger and more efficient with each new node, operating transparently without censorship. Nikitin explained, “As more nodes join, our network’s intelligence expands, creating a planetary-scale AI reasoning model built on community-owned models. These models collaborate and improve continuously through swarm inference, ultimately surpassing centralised AI in reasoning and efficiency.”
The platform allows anyone to run a node, with ratings recorded on the Monad blockchain for evaluation and security. Developers can access frontier AI reasoning via API or on-chain, while node operators receive rewards for their contributions.
Nikitin emphasised: “Fortytwo pioneers Swarm Inference as a fundamentally new approach to AI scaling. Instead of one large model, requests are handled by specialised models working together on everyday PCs and Macs. The network peer-reviews responses to prevent errors and hallucinations. Each model contributes specific expertise, achieving superior task-specific results.”
Unlike centralised AI, Fortytwo grows without costly centralised training. Each new node and model strengthens the network, making it intelligence community-driven. Unlike inference marketplaces, Fortytwo’s swarm inference combines multiple models, creating intelligence surpassing any model’s capabilities. This makes AI more cost-efficient, scalable, and resilient, improving accuracy as the network expands.
Regarding competition, Nikitin noted: “Fortytwo competes with centralised AI companies like OpenAI, Anthropic and DeepSeek. However, we’re not just another LLM provider — we represent a new paradigm in AI inference, offering decentralised community-owned intelligence as an alternative to traditional, capital-intensive AI infrastructures.”
Pioneering a new era of distributed AI infrastructure
Sam Kim, Partner at Big Brain Holdings, commented: “We invest in research-driven teams that take a deep technical approach to solving fundamental AI challenges. As centralised AI reaches its limits, Fortytwo is leading the shift to a decentralised, globally distributed future, and we’re excited to support their work.”
Nikitin concluded: “Our long-term vision is to build the world’s first permissionless, community-driven planetary-scale reasoning engine — a decentralised intelligence network open to all contributors. We aim to establish Fortytwo as the ‘Wikipedia of inference’ — an open system where individuals and organisations can contribute specialised AI models while maintaining ownership and data privacy. This ensures AI remains a shared, accessible resource, not confined to a few centralised corporations.”
The post Exclusive: Peer-to-peer AI network Fortytwo grabs $2.3M to take on OpenAI, Anthropic and DeepSeek dominance appeared first on Tech Funding News.