IBM and Together AI sign $240 million AI infrastructure deal
IBM and Together AI have agreed to a multi-year $240 million deal to build an Nvidia-powered AI inference cluster on IBM Cloud.
IBM and Together AI have signed a multi-year deal worth $240 million to build a large-scale AI inference cluster on IBM Cloud. Reported on 11 August 2026, the agreement centres on Nvidia-powered infrastructure designed to help businesses run AI models faster and at greater scale.
The cluster will use Nvidia HGX B300 systems and Nvidia Spectrum-X networking, giving Together AI access to high-performance computing capacity through IBM's cloud platform.
What the deal includes
The centrepiece of the agreement is an AI inference cluster. In simple terms, inference is the stage when an AI model is used to produce an output after it has been trained. When a chatbot answers a question, a coding assistant generates code, or an AI model processes a request, that is inference.
As businesses move AI from experiments into everyday operations, demand for inference computing is increasing. Training large models requires substantial resources, but serving those models to millions of users can require even more computing capacity over time.
The Nvidia HGX B300 systems in the deal are designed for demanding AI workloads. Spectrum-X networking is built to move data efficiently between systems, allowing large computing clusters to operate as connected environments.
Why IBM is betting on AI infrastructure
IBM has been building its enterprise AI strategy around hybrid cloud, specialised infrastructure and regulated industries. The company has also expanded its relationship with Nvidia, bringing newer GPU technology to IBM Cloud for AI training, reasoning and inference workloads.
The Together AI agreement gives IBM another opportunity to demonstrate that its cloud infrastructure can support large AI deployments. IBM does not have the same public cloud market share as Amazon, Microsoft or Google.
However, its long-standing relationships with enterprises in sectors such as financial services, healthcare and government give it an established customer base for AI services. A partnership with Together AI could help IBM attract organisations looking for scalable AI computing without having to build and operate their own infrastructure.
What Together AI gains
Together AI provides infrastructure and tools for companies developing and deploying generative AI models, including open-source models. The agreement gives the company access to a dedicated Nvidia-powered cluster through IBM Cloud.
This can help it support customers that need reliable computing capacity for high-volume inference. Access to advanced GPUs has become a strategic issue across the AI industry. Demand remains strong as companies need computing resources not only to train models but also to fine-tune and operate them.
A multi-year infrastructure agreement can give Together AI greater capacity and predictability as it expands its services.
Why inference is becoming so important
The AI infrastructure race is increasingly moving beyond model training. Once an AI system is deployed, every user request consumes computing resources. As AI assistants, coding tools, enterprise copilots and automated agents become more common, the amount of inference required can grow rapidly.
That makes speed, availability and cost increasingly important. Companies need infrastructure that can handle large workloads without creating excessive delays or operating costs. For cloud providers and AI infrastructure companies, this creates an opportunity to differentiate themselves through specialised hardware, networking and efficient model serving.
What the deal means for the AI market
The $240 million agreement reflects a wider change in the industry. The focus is no longer only on building larger and more capable AI models. It is also about running those models efficiently, securely and reliably in real-world environments.
For enterprises, partnerships such as this could make advanced AI infrastructure easier to access without requiring direct investment in data centres and specialised hardware.
For IBM, the deal supports its ambition to become a major infrastructure provider for enterprise AI. For Together AI, it provides additional capacity at a time when inference is becoming one of the most important parts of the AI economy.
The deal also highlights Nvidia's continued influence. Even as cloud providers develop alternatives and custom accelerators, Nvidia GPUs and networking technologies remain central to many large-scale AI deployments.
A sign that enterprise AI is scaling
IBM and Together AI's $240 million agreement is more than a cloud infrastructure contract. It shows that AI deployment is moving from experimentation towards large-scale production. As businesses demand faster and more dependable AI services, partnerships combining cloud platforms, specialist AI providers and advanced chips are likely to become increasingly important.