Fireworks AI
"Own Your Specialized Intelligence"
Redwood City AI infra startup (2022) serving open-source LLMs like Kimi K3 via API, used by Cursor, Vercel and Notion.
Fireworks AI is a generative-AI infrastructure company founded in 2022 and headquartered in Redwood City, California. It was started by a team of seven engineers with backgrounds building and maintaining PyTorch and large-scale machine learning infrastructure at Meta and Google, including CEO Lin Qiao, formerly head of PyTorch at Meta. The company operates a training and inference platform that lets organizations deploy, fine-tune and serve open-source large language models -- including DeepSeek, GLM, Kimi, Qwen and Minimax -- through OpenAI- and Anthropic-compatible APIs. Its inference offering spans serverless pay-per-token access (priced from roughly $0.07 to $3 per million tokens depending on model and priority tier), on-demand dedicated deployments, and reserved capacity for high-volume workloads, running across multiple cloud regions on NVIDIA and AMD hardware. On the training side, Fireworks AI supports LoRA and full fine-tuning, guided training runs with cost estimation, and integrated reinforcement-learning pipelines for building custom or task-specific models. In August 2026, Fireworks AI began hosting Moonshot AI's Kimi K3 model on Microsoft Foundry, extending its role as a third-party inference backend embedded in a major cloud AI marketplace. Companies publicly named as customers on the Fireworks AI website include Cursor, Vercel, Notion, UiPath, Sourcegraph, Genspark and Motif. The company has raised venture funding from Benchmark, Sequoia Capital, Lightspeed Venture Partners and Index Ventures, with NVIDIA and AMD also participating as strategic infrastructure investors, and has been reported at a multi-billion-dollar valuation following funding rounds through 2025. According to a 2026 employer-profile snapshot, Fireworks AI has approximately 63 employees, a relatively lean headcount consistent with an infrastructure-focused engineering organization rather than a large services business. Fireworks AI is not a robotics hardware manufacturer; it is a horizontal AI model-serving and fine-tuning platform. Its relevance to the robotics and embodied-AI space comes from providing low-latency inference and custom-model training infrastructure that downstream developers -- including those building agentic software, multimodal systems and robotics-adjacent AI applications -- can use to deploy foundation models in production without operating their own GPU clusters.
Primary type & automation activities this supplier delivers:
Fireworks Inference Platform
Serverless and dedicated LLM inference, fine-tuning and multi-model routing (Fireworks Nexus) platform hosting open-source models such as DeepSeek, Kimi and Qwen.
Contact Fireworks AI
WEBSITE
https://fireworks.aiHEADQUARTERS
United States
Company Facts
Founded
2022
Primary Role
Software/Algorithm
Company Size
employees 50-100
Primary Region
North America
Annual Sales
-
Funding Stage
-
Funding Total
-
CASIA's Peiyan Li Brings Two VLA Papers to IROS 2026
CASIA Ph.D. candidate Peiyan Li co-authored two IROS 2026 papers on VLA models for long robot tasks. We also cover his BridgeVLA and SpatialVAM work.
LG Electronics Showcases Robotics at 2026 First Global Challenge in Incheon
LG Electronics is actively participating in the 2026 First Global Challenge, an international youth robotics competition taking place at Songdo Convensia in Incheon. The event, which runs from Wednesday to Saturday, gathers youth teams from 185 countries to design and build robots aimed at addressing global challenges, particularly focusing on wildfire prevention and recovery due to climate change. This year's competition, themed “Igniting Innovation,” emphasizes the importance of robotics in tackling pressing issues like wildfires. LG's involvement includes sponsoring the event and showcasing its latest robotics technologies at an exhibition booth. The company also features its humanoid robot, LG CLOiD, allowing visitors to engage with the technology and learn about the innovative solutions being developed by young participants. As LG increases its investment in robotics as a key growth area, it aims to integrate physical AI into its strategy. The company recently launched LG Data Factory to enhance data management for robotics training. No further timeline was disclosed at the time of publication.
All News
Zhiyuan's Data Business Struggles to Support Robotics at $20 Per Hour
Zhiyuan's data business is facing challenges as it operates at a rate of $20 per hour, which is insufficient to sustain the robotics industry. The company is exploring new avenues to enhance its offerings and address the growing demand for data in robotics. This situation highlights the broader issue within the robotics sector, where the cost of data collection and processing is becoming a critical factor. As companies seek to innovate and improve their robotic systems, the financial viability of data services like those offered by Zhiyuan is under scrutiny. Looking ahead, industry stakeholders should monitor how Zhiyuan adapts its business model to better serve the robotics market. No further timeline was disclosed at the time of publication.
Robotics Automation AI
Fudan Team Develops Robot Mat Capable of Harvesting Ground Electricity, Eliminating Battery Needs
A team from Fudan University has developed an innovative robot mat that can harness electricity from the ground, thus removing the necessity for batteries. This advancement could significantly impact the design and functionality of mobile robots, making them more efficient and environmentally friendly. The ability to extract energy directly from the ground is a game-changer in robotics, as it addresses one of the major limitations of mobile robots: battery dependency. This technology could lead to longer operational times and reduced maintenance costs, enhancing the overall performance of robotic systems in various applications. Future developments in this area will be crucial to watch, particularly how this technology can be integrated into existing robotic platforms. No further timeline was disclosed at the time of publication.
Robotics Automation AI
Huawei Unveils XMFS Prototype for Cross-Node Shared Memory at LPC 2026
Huawei presented its XMFS prototype at the Linux Plumbers Conference 2026, showcasing a new approach to cross-node shared memory. This innovative system utilizes POSIX file calls to facilitate memory sharing across different nodes, enhancing performance and efficiency in distributed computing environments. The introduction of XMFS is significant as it addresses the growing need for efficient memory management in multi-node systems. By enabling cross-node shared memory, Huawei aims to improve data accessibility and processing speed, which are critical for applications requiring high-performance computing. Looking ahead, industry observers will be keen to see how XMFS evolves and whether it will undergo further development to achieve mainline review status. The potential impact of this technology on distributed computing and memory management practices will be closely monitored in the coming years.
New Mobile System Enhances Drone Operations with Persistent Nanogrid Technology
Three US companies, Sesame Solar, Target Arm, and AG3 Labs, have introduced the first Made-in-USA Persistent Drone Operations Nanogrid. This innovative mobile system integrates power generation, autonomous drone launch and recovery, and configurable small drones to enhance operational efficiency in combat and other environments. The significance of this development lies in its potential to minimize downtime for drones, which is crucial for maintaining continuous operations. By automating the launch, recovery, and recharging processes, the Nanogrid addresses challenges faced by drone teams, particularly in remote or contested areas where conventional fuel resupply is difficult. Looking ahead, the companies anticipate that this system will find applications beyond military use, including border monitoring, search and rescue, and critical infrastructure inspections. No further timeline was disclosed at the time of publication.
News
Xiaomi Expands Apple Device Integration with HyperOS 4 Features
On October 7, Xiaomi announced enhanced integration with Apple devices in HyperOS 4, allowing users to share photos, files, contacts, and web pages seamlessly across both ecosystems. This new functionality includes two-way access to iPhone and Xiaomi cloud photo albums, notification syncing, and the ability to reply to messages without switching devices. The integration supports various Apple devices, including iPhone, iPad, and Mac. Users can view and control their Xiaomi phone through Xiaomi’s connectivity app on Mac and iPad, provided they have compatible software versions, a shared Xiaomi account, and the necessary network connection. This feature aims to enhance user experience by bridging the gap between Xiaomi and Apple ecosystems. HyperOS 4 was initially introduced with the Xiaomi 18 Pro series in September, and a wider rollout is set to commence on October 15. As Xiaomi continues to expand its software capabilities, it will be interesting to observe how this integration impacts user adoption and the competitive landscape between Xiaomi and Apple.
News Feed
Mecka Secures $60 Million in Series B Funding to Enhance Robotics Data Infrastructure
Mecka has successfully raised $60 million in Series B funding, led by Sequoia Capital, to enhance its robotics data infrastructure and commercial deployments. The company specializes in capturing human demonstrations using proprietary multi-sensor hardware, converting this data into training signals for robotic systems, and collaborating with researchers on the EgoVerse study. This funding is significant as it positions Mecka to bridge the gap between available AI training data and the real-world operational needs of robots. With a reported run-rate revenue exceeding $100 million as of June 2026, Mecka anticipates reaching $300 million by year-end, indicating strong growth in the robotics sector. Looking ahead, Mecka plans to utilize the new capital to develop more data-capture hardware and expand its research and deployment capabilities. The company aims to provide tailored training data for various robotic applications, emphasizing the importance of human-like learning for future robotic systems.
AI Funding & Investment Physical AI Startups AI Data deployment
IROS 2026 Best Paper Finalists: South Korea Leads
Ten papers made the IROS 2026 Best Paper and Best Student Paper shortlist, four from South Korea. What each one does and what it means for industry.
Microsoft Unveils New AI PCs Featuring Nvidia's RTX Spark Chip and Enhanced Windows 11
Microsoft has announced the specifications and pricing for its new Surface Laptop Ultra and Surface RTX Spark Dev Box, powered by Nvidia's RTX Spark chip. The Surface Laptop Ultra starts at $2,600, while the Surface RTX Spark Dev Box is priced from $6,000, designed for on-device AI model execution with enhanced hardware capabilities. These new devices are significant as they cater specifically to developers, offering advanced features like superior cooling and a revamped Windows 11 with 'Execution Containers' for easier AI agent sandboxing. Microsoft CEO Satya Nadella emphasized the need for orchestration and memory management in AI applications, highlighting the platform's future direction. Looking ahead, Microsoft is positioning these PCs as versatile tools not only for developers but also for content creation and gaming. The company is incentivizing trade-ins of MacBook Pros, reflecting the competitive landscape in the developer community. No further timeline was disclosed at the time of publication.
AI Hardware TC Microsoft nvidia
SimScale Engineering AI Moves Physics Simulation to the Exact Moment a Design Decision Gets Made, Inside Onshape
SimScale’s Engineering AI Agent runs CFD, thermal, EMag and FEA from one prompt inside PTC’s Onshape, starting from CAD that isn’t simulation-ready.
IMTS 2026 Preview: Record Orders and the Automation Story
IMTS 2026 opens in Chicago with 1,788 exhibitors and record H1 machine tool orders. What the USMTO data says about automation, and the booths to see.
Software, Compute and Interaction at WRC 2026
47 WRC 2026 exhibitors supply robot software, compute or interaction — but only 19 call themselves software companies. Five open artefacts, licence-checked.
WRC 2026 Exhibitor Map: 248 Robotics Companies Mapped
We classified 248 WRC 2026 exhibitors with the RobotToday taxonomy. Motion & Actuation leads at 75 companies, Humanoid at 72, and compute stays thin at 12. (Modified 2026-08-27)
Inside Saga Robotics: UVC Light vs. Crop Disease
Saga Robotics CPO Damian Flynn on the Thorvald platform's UVC treatment system, its jump from UK strawberries to California wine grapes, and why he's “incredibly bullish” on applying automation and robotics into industries “most chaotic environment.”
WAIC Review 4/4: From Port AGVs to Hospital Night Shifts: What WAIC 2026 Reveals About Real Robot Deployments
From 530+ AGVs at a Singapore port to robots on Apple's line, WAIC 2026 speakers shared concrete data on where embodied AI robots are actually working today.
WAIC Review 2/4: World Models vs. VLA: Robotics' Next 'ChatGPT Moment'
At WAIC 2026, world-model and VLA camps staked rival paths to generalist robots; Zhiyuan, SenseTime and NVIDIA predict a robot 'ChatGPT moment' in 2-5 years.
Autonomous Mobile Robots Siegwart, Nourbakhsh & Scaramuzza — RSF Top 10 Robot Education Textbooks · No. 3
AMR and AGV fleets are the biggest robot service market. Siegwart decodes the vendor manual: odometry, localization, cost maps.
Moonshot AI's Kimi K3 Now Accessible to Enterprises via Fireworks on Microsoft Foundry
Moonshot AI has made its Kimi K3 model available through Fireworks AI on Microsoft's Foundry platform, enabling Azure customers to utilize the model via OpenAI-compatible APIs. This integration allows companies to transition open models from evaluation to production without the need for managing their own GPU infrastructure. The significance of this development lies in the streamlined access it provides to enterprise users, facilitating the deployment of AI models in a managed environment. With Kimi K3 listed among over 20 open models on Foundry, organizations can leverage the capabilities of Fireworks AI's inference engine alongside Microsoft's governance and billing features. Looking ahead, the focus will be on how enterprises adopt the Kimi K3 model within their operations. As Moonshot AI continues to expand its distribution channels, the emphasis will be on enhancing the accessibility and usability of AI solutions for businesses. No further timeline was disclosed at the time of publication.
News Feed
New Data Shows NVIDIA Blackwell Ultra Delivers up to 50x Better Performance and 35x Lower Costs for Agentic AI
The NVIDIA Blackwell platform has gained significant traction among top inference providers, including Baseten, DeepInfra, Fireworks AI, and Together AI, achieving reductions in cost per token by as much as 10 times. Building on this success, NVIDIA has introduced the Blackwell Ultra platform, which aims to enhance these cost efficiencies further. This development reflects NVIDIA's commitment to advancing AI technology and providing more affordable solutions for inference tasks, thereby supporting the growing demand for cost-effective AI applications in various industries.
