IBM launched Watsonx, a platform for artificial intelligence models and generative AI, in a move that revives the Watson brand. IBM also announced a partnership with Hugging Face to bring open source AI models to the enterprise.
Last week, ServiceNow and Hugging Face announced a partnership on open source AI models.
For IBM, the Watsonx launch is a way to bring together foundational models for code, AIOps, digital labor, security and sustainability. IBM is looking to capitalize on the generative AI boom as well as its various use cases.
IBM announced the news at its Think conference. IBM said it is looking to provide a full enterprise stack to train, tune and deploy AI models. In addition, IBM is outlining GPU-as-a-service to support AI workloads.
Watsonx includes the following:
- An AI development studio to access curated and trained open source models. The AI studio will also enable enterprises to manage the data and AI lifecycle.
- The ability to build models on language as well as code, time-series and tabular data, geospatial data and IT events.
- Access to a data store to train and cleanse data via IBM Watsonx.data, which is built on an open data lakehouse architecture. It will be generally available in July and manage workloads on-premise and in multi-cloud environments.
- A set of governance tools via IBM Watsonx.governance, which aims to govern customer privacy, model bias and drift as well as provide transparency.
- Hugging Face will contribute open-source libraries, models and datasets. IBM is looking to use the open source approach it has used historically for AI applications.
IBM will also include a Watson Code Assistant, AIOps Insights, Watson Assistant and Orchestrate and Environmental Intelligence Suite as part of the Watsonx foundation.
Here's Constellation Research analyst Andy Thurai's take:
As with many announcements lately, IBM has YAGAIA (Yet another generative ai announcement). It is sad to say this time it is the granddaddy of AI, IBM.
Among the WatsonX announcements, AI studio and datastore won't move the needle much. There are other vendors who are offering better AI data stores data lakes for many years. However, the governance toolkit is an interesting option. This allows companies to build trusted AI workflows and help with building explainable and transparent AI workflows. Though it is not available currently, slated to be released late this year, IBM can be differentiated in the crowded AI market.
As far as the Hugging Face relationship is concerned, it is more of a win for Hugging Face than for IBM. Hugging Face has become the de facto standard for public model repositories. They have partnerships with many recently among the same lines - AWS, ServiceNow, and even Azure. Most of these companies don't want to compete in the LLM arms race instead they want to enable enterprises to train their own models by providing the necessary components and providing a way those models can be easily created, shared, and monetized across the marketplace.
IBM’s GPU as a service is very late to the market. Many Cloud and HPC vendors have been offering this for years. While it is not a game changer, it is good to see that IBM is catching up in this area. This is a much similar story to IBM entering the cloud market very late when the other players had already established and matured their services. Their AI-optimized infrastructure is still not available, yet, which is scheduled for a full release later this year. While the efficiency in itself needs to be proven, it is going to be hard to win against other players in this area.
IBM’s Center of Excellence of AI experts is a brilliant idea. This is one of the areas where most of the other AI companies struggle - to create an army of practitioners to help customers. IBM Consulting has been in front of customers helping major enterprises with digital transformation for decades. This practice, with over 1,000 experts, who are focused on WatsonX-practices can be developer advocates in creating traction for the technology with the customers. While this is still just in the announcement stage, the move can put IBM in the right AI direction.
The Environmental Intelligence Suite, which helps calculate environmental risks and provides a carbon calculator can be a unique offering for companies that care about the environment. However, no one in the AI space is worried too much about emissions, carbon footprint, or even efficiency. Fast to market and AI washing everything at a faster pace than others seem to be the norm now.
Overall, IBM had a good set of announcements, mostly catching up with others. I still need to see the execution to believe this is all real.