What’s New: Today, Inflection AI and Intel announced a collaboration to accelerate the adoption and impact of AI for enterprises as well as developers. Inflection AI is launching Inflection for Enterprise, an industry-first, enterprise-grade AI system powered by Intel® Gaudi® and Intel® Tiber™ AI Cloud (AI Cloud), to deliver empathetic, conversational, employee-friendly AI capabilities and provide the control, customization and scalability required for complex, large-scale deployments. This system is available presently through the AI Cloud and will be shipping to customers as an industry-first AI appliance powered by Gaudi 3 in Q1 2025.
"Through this strategic collaboration with Inflection AI, we are setting a new standard with AI solutions that deliver immediate, high-impact results. With support for open-source models, tools, and competitive performance per watt, Intel Gaudi 3 solutions make deploying GenAI accessible, affordable, and efficient for enterprises of any size."
Why It Matters: Building an AI system typically demands substantial infrastructure--extensive model development and training, and collaboration among engineers, data scientists and application developers. With Inflection for Enterprise, built on Inflection 3.0, enterprise customers can now harness a comprehensive AI solution that empowers their workforce with a virtual AI co-worker specifically trained on their unique company data, policies and culture. The partnership with Intel brings unmatched performance through the Intel Gaudi 3 AI accelerator, which offers industry-leading price/performance for efficient, high impact results. Intel’s technology ensures flexibility and scalability for high-impact results. Additionally, the AI Cloud streamlines the building, testing and deployment of AI applications in a unified environment, accelerating time to market. With the value and benefits this service offers, Intel and Inflection AI are also collaborating to deploy Inflection for Enterprise within Intel with the anticipation that Intel will be an early customer of the solution.
“Every CEO and CTO we speak to is frustrated that existing AI tools on the market aren’t truly enterprise-grade,” said Inflection AI COO Ted Shelton. “Enterprise organizations need more than generic off-the-shelf AI, but they don’t have the expertise to fine-tune a model themselves. We’re proud to offer an AI system that solves these problems, and with the performance gains we see from running on Intel Gaudi, we know it can scale to meet the needs of any enterprise.”
How It Works: Inflection AI fine-tunes its model to be native to each organization, expediting user adoption and improving the usefulness of use cases through alignment with the company’s tone, purpose, and unique product, service, and operating information. Inflection 3.0 enables enterprise customers with faster time-to-value through employee-friendly generative AI experiences, while offering price, performance and security/compliance advantages.
- Removing Barriers to GenAI – Built on AI Cloud, Inflection for Enterprise provides application templates designed to let businesses skip hardware testing and model building and avoid capital expenses to scale quickly. In Q1 2025 customers will also have the option to purchase Inflection for Enterprise on a complete turnkey AI appliance. Leveraging Gaudi 3, customers of this appliance can benefit from up to 2x improved price performance as well as 128GB of high-bandwidth memory capacity further optimizing their GenAI performance compared with current competitive offerings.
- Optimized Price/Performance – While Inflection AI’s Pi consumer application was previously run on Nvidia GPUs, Inflection 3.0 will be powered by Gaudi 3 with instances on-premises or in the cloud powered by AI Cloud. This not only cuts down on time to deploy but also total cost of ownership.
- Fine-Tuned for Enterprises – Leveraging the fine-tuning and reinforcement learning from human feedback (RLHF) expertise that powered Inflection AI’s Pi, Inflection for Enterprise models are unique to each business’ ethos and way of operating. Modeled on data and insights from a company’s history, policies, content, tone, products and operating information, Inflection AI helps drive productivity and alignment across an organization.
- Enhanced Ownership and Security – Inflection for Enterprise allows enterprises to own their intelligence in its entirety. Fine-tuned models are the customer’s alone and are never shared outside their organization. Additionally, customers can host and run the model on their preferred architecture, whether hosted on-premises, in the cloud, or hybrid.
What’s Next: Looking ahead, Inflection AI and Intel will also enable developers to build enterprise applications for Inflection for Enterprise, leveraging the robust and human-centric Inflection 3.0 system, to generate critical software tools. Enterprise customers interested in Inflection for Enterprise, please visit https://inflection.ai/intel to learn more or sign up for a demo.
Source: Intel measured results vs H100 data sources: https://github.com/NVIDIA/TensorRT-LLM/blob/main/docs/source/performance/perf-overview.md input-output
Sequences: 128-2048tps on 2 accelerators/GPUs. Intel results obtained on September 9th 2024.
Hardware: Two Intel Gaudi 3 AI Accelerators (128 GB HBM) vs two Nvidia H100 GPU (80 GB HBM).
Software: Intel Gaudi software release 1.18.0.
See Nvidia link for H100 software details.
Results may vary.
Pricing estimates based on publicly available information and Intel internal analysis