The former Apple Engineer, Neil Movva, founded by Sail Research, ended invisibility and completed $80 million in seed wheels and round A finance, valued at $450 million. The company is not targeting chat robots, but is running for several hours in a sustainable manner, while also dealing with a large number of mission-driven AI intelligence reasoning needs.
Financing and positioning
Kleiner Perkins leads the Sail A round, Sequoia, Redpoint, Theory Ventures, Vine Ventures and CRVs participate. According to Fortune, the company wants to address the rapid increase in the bill of reasoning after the company deploys AI intelligence.
Long-term task push costs
Most of the existing AI infrastructure is structured around low-delayed design, with a focus on returning results as soon as possible and suitable for chatting and real-time interaction. However, the enterprise is using AI for tasks of a longer process, such as reading the complete code library, mass screening of candidates, or completing complex studies without intervention.
The report cites corporate accounts that the token consumption of this type of intelligent workflow can be 50 to 500 times as high as the simple conversation scene. Goldman Sachs predicted that by 2030, token consumption could increase 24 times, which also made reasoning efficiency a new point of competition.
In exchange for efficiency.
The Sail approach is not to develop a self-study chip, but to develop a schedule and optimization software around the existing GPU to allow the model to perform more computing tasks on the same hardware. Unlike most reasoning platforms that prioritize low delays, Sail places more emphasis on throughput and resource utilization.
- Total financing US$ 80 million
- Corporate valuation reached $450 million
- Client costs have improved approximately 3 to 10 times
Movva indicates that this trade-off is not appropriate for voice assistants or real-time chats, but is more appropriate for a smart body mission that continues for hours. According to the company, clients are generally able to obtain a more visible cost reduction in such situations.
March online reasoning service
Movva worked in NVIDIA, Apple and Together AI. Sail co-founder and CTO Samir Menon also came from Apple. They re-engaged at the end of 2025 and began to rebuild the reasoning post from the ground floor.
According to the company, Sail had introduced the reasoning service in March of this year, and the number of tokens being processed each week had reached a trillion. Early client Detail.dev uses the platform's running code to review intelligence, which usually takes 3-4 hours, or even longer.
Additional information:The report mentions that Sail faces competition not only from Together AI, but also from head-model companies like Anthropic, OpenAI and Google, which are building their own reasoning infrastructure.