Thinking Machines Lab and Crusader Cloud have signed a $65 million‑per‑year agreement that places Crusader’s managed inference service at the core of running the lab’s open‑source models. The deal, announced by both parties, formalizes a long‑term collaboration between the AI research outfit founded by former OpenAI CTO Mira Murati and the cloud provider that specializes in inference workloads.
Key Takeaways
- The partnership is valued at $65 million each year.
- Crusader Cloud will operate inference for Thinking Machines Lab’s open models.
- Thinking Machines Lab was founded by Mira Murati, former OpenAI CTO.
- Myle Ott leads machine‑learning infrastructure at the lab.
- Erwan Menard serves as senior vice‑president of Crusader Cloud.
| Metric | Value | Why it matters |
|---|---|---|
| Annual contract value | $65 million | Signals a sizable revenue stream for Crusader and a dedicated inference pipeline for the lab. |
| Partnering companies | Thinking Machines Lab & Crusader Cloud | Combines cutting‑edge research with production‑grade inference infrastructure. |
| Founder of Thinking Machines Lab | Mira Murati | Her OpenAI background lends credibility and technical depth to the lab’s work. |
| ML infrastructure lead | Myle Ott | Oversees the engineering stack that will integrate Crusader’s service. |
| Senior vice‑president | Erwan Menard | Represents Crusader’s executive commitment to the partnership. |
The deal in detail
The agreement designates Crusader Cloud as the exclusive provider of managed inference for the open‑source models released by Thinking Machines Lab. By leveraging Crusader’s “Managed Inference” platform, the lab can offload latency‑critical serving workloads to a cloud environment optimized for GPU‑accelerated inference, while retaining control over model architecture and licensing.
Thinking Machines Lab: research pedigree
Founded by Mira Murati, a former chief technology officer at OpenAI, Thinking Machines Lab positions itself as a research‑first organization that publishes large language models (LLMs) under permissive licenses. The lab’s focus on openness aims to democratize access to state‑of‑the‑art AI while encouraging community‑driven improvements.
Myle Ott, the lab’s ML infrastructure lead, is tasked with ensuring that the inference pipeline scales efficiently. His role bridges the gap between experimental model development and production‑grade serving, a critical function given the computational intensity of modern LLMs.
Crusader Cloud’s managed inference offering
Crusader Cloud markets a service called “Crusader Managed Inference,” which abstracts away the complexities of GPU provisioning, autoscaling, and low‑latency networking. Clients submit model artifacts via an API, and the platform handles request routing, batching, and hardware selection automatically.
Senior vice‑president Erwan Menard highlighted the strategic fit of the partnership, noting that the lab’s open models provide a steady workload stream that aligns with Crusader’s focus on high‑throughput inference workloads.
Technical implications for model serving
Running inference at scale typically requires a combination of high‑bandwidth interconnects, tensor‑core GPUs, and sophisticated software stacks such as NVIDIA Triton or custom SDKs. Crusader’s managed service promises to deliver these capabilities without the end‑user needing to manage the underlying hardware, reducing operational overhead for Thinking Machines Lab.
The partnership also opens the door for the lab’s community to experiment with real‑world latency and cost metrics, potentially informing future model design decisions. By centralizing inference on Crusader’s platform, the lab can collect telemetry that feeds back into research cycles.
Context within the AI inference market
The AI inference market has seen a surge of specialized cloud providers offering turnkey solutions for LLM serving. The $65 million annual commitment underscores the growing willingness of research labs to allocate substantial budgets toward production‑grade inference, reflecting a broader industry shift from pure research to commercializable AI services.
Why it matters
This agreement illustrates how open‑source AI initiatives are beginning to monetize at scale through dedicated inference infrastructure. For developers, the deal promises more reliable access to cutting‑edge models without the need to manage expensive GPU clusters. For investors and industry watchers, the $65 million figure signals that cloud providers can extract significant value from serving open models, potentially reshaping revenue models for AI research labs.
Frequently asked questions
Who signed the $65 million‑per‑year agreement?
Thinking Machines Lab and Crusader Cloud are the two parties that entered into the contract.
What does the agreement cover?
Crusader Cloud will provide its Managed Inference service to run inference on the open models released by Thinking Machines Lab.
Who founded Thinking Machines Lab?
The lab was founded by Mira Murati, who previously served as chief technology officer at OpenAI.
What role does Crusader Managed Inference play?
It offers a turnkey platform that handles GPU provisioning, autoscaling, and low‑latency request routing for AI models, allowing the lab to focus on research rather than infrastructure.
Which executives are highlighted in the announcement?
Myle Ott is identified as the ML infrastructure lead at Thinking Machines Lab, and Erwan Menard is named senior vice‑president at Crusader Cloud.
Bottom Line
The $65 million‑per‑year partnership between Thinking Machines Lab and Crusader Cloud formalizes a substantial investment in managed inference, pairing a research‑driven open‑model ecosystem with a cloud service built for high‑throughput AI serving. The deal not only secures a reliable inference pipeline for the lab’s models but also highlights the commercial potential of open AI in the cloud‑inference market.
We will watch how the collaboration influences model performance metrics and whether similar agreements become a norm for other open‑source AI initiatives.
Reporting based on: original source.
Comments
Post a Comment