News Analysis
IBM dusted off its Watson brand for the generative AI era this week with the roll out of its Watsonx AI and data platform. The launch follows a beta testing period in which 150 enterprises participated.
The company first previewed the platform at IBM Think in May. And while a big tech firm jumping into the generative AI race is de riguer at this point, the combination of IBM's long history in AI and the failure of Watson ever to find its niche leave many wondering if Big Blue can succeed in resuscitating the brand.
IBM Goes All In on Generative AI With Watsonx
Unlike many of the other generative AI releases of the past six months, Watsonx tries to cover the entire range of enterprise needs across three products:
- Watsonx.ai studio: Offers new foundation models, generative AI and machine learning.
- Watsonx.data: This fit-for-purpose data store is built on an open lakehouse architecture.
- Watsonx.governance: A toolkit to help businesses build AI workflows with responsibility, transparency and explainability.
The company made the first two available as of July 11, with governance expected later this year. Future releases will also provide access to a greater variety of IBM-trained proprietary foundation models for efficient domain and task specialization.
The release came with two other significant developments, one being the availability for AI builders to use models form the Hugging Face community for a range of AI tasks.
Founded in 2016, Hugging Face develops tools for building applications using machine learning. It is most notable for its transformers library, built for natural language processing applications, and its platform that allows users to share machine learning models and datasets.
The models offered by Hugging Face are pre-trained to support a range of natural language processing (NLP) type tasks including question answering, content generation and summarization, text classification and extraction.
Second comes the news first reported by Reuters that IBM is considering using its own proprietary AI chips to reduce the costs of operating Watsonx. The chips, called the Artificial Intelligence Unit, were first mentioned in October 2022. According to the report, the Samsung Electonics manufactured chips would operate in conjunction with Nvidia chips, focused specifically on "inference, which is the process of putting an already trained AI system to use making real-world decisions."
Related Article: What Dolly 2.0 Means for the Digital Workplace
The Challenges Ahead for IBM
From a global perspective, Watsonx could be a game changer for enterprise AI, particularly in a world of LLMs and generative AI, said Alan Pelz-Sharpe, founder and principal analyst at Deep Analysis. It brings much-needed structure, governance and the possibility of building trusted and accountable systems.
The challenge for IBM, however, is twofold, he continued:
- The Watson brand is tainted after years of making and failing to deliver on grandiose promises (Watson Health, for example).
- The challenge of making enough money from Watsonx to make it a viable business for IBM.
In an AI world flooded with open-source alternatives, Pelz-Sharpe said, the vendor's best bet is selling associated services to build generative AI systems for its most prominent clients.
“All that said, it’s a worthy announcement; the AI world at the moment is chaotic, full of promise but also full of regulatory concerns and valid doubts about its risks, accuracy and enterprise value,” he said.
How Far Can Watsonx Go?
IBM has a long track record in AI, but Watsonx could really shake things up, if not now then soon, said Gartner distinguished VP and analyst Arun Chandrasekaran. If Watsonx is not a market disruptor yet, it will provide incremental growth opportunities for IBM — provided they can execute on their vision well, he continued.
However, the risks of deploying generative AI applications are “non-trivial” for enterprises. “Hence, vendors that can demonstrate superiority across the above principles stand to benefit,” he said.
In respect to the Hugging Face partnership, he points out that it partners with other entities beyond IBM (such as AWS). IBM as a result, he said, should reinforce its commitment to open source in its own models and build operational and responsible AI tools for Hugging Face models to persuade customers to go along with them.
He added that Watson will be a brand aimed at business function centered products (IT, Customer Service, HR among others) while Watsonx is aimed at builders. “IBM needs to accelerate its research to product cycle and forge strong partnerships with ISVs and consulting firms to gain mindshare in Generative AI," he said.
Related Article: How Generative AI May Fit Into Your Organization
The Watson Legacy
The Watson brand has a mixed legacy in bringing AI to market, said Matt Mullen, lead analyst, AI applications at Deep Analysis.
“It certainly has some remaining resonance around the whole Jeopardy business but that does rather stand as a reminder of how long IBM has been pushing this towards enterprises or adoption on a raft of so far, undelivered promises, and worse with regard to Watson Health of course,” he said.
Once you move past the excitement being generated by the broad promise of generative AI, quite quickly you arrive at the "so what?" question, he continued.
The enterprise application vendors bringing similar sets of AI capabilities to their suites should look to the processing points within existing workflows to act as a route-in for the technology for their customers, Mullen suggested. “Standalone, as Watsonx is here, then you're relying on customers to do that application specific work for you,” he said.
He said the real audience here is integrators, SIs and ISVs looking for new value additions at a reasonable human cost (i.e., not requiring as expensive a set of skills), rather than a situation where even technically adept end users will be picking up toolkits like this and immediately running productive projects.
As for the Hugging Face addition, he argues that in marketing terms right now this is an important offering but says that the practical reality is that only a small number of big foundation models will be used by most people, most of the time.
“The specialist models — maybe even customer specific ones — are where the real multiplicity will lie, and we're not there yet,” he said. “While the development process is easier — and has been since transformer models appeared 5-6 years ago [a transformer model is a neural network that learns context and, consequent meaning by tracking relationships in sequential data] — the skills to build them are still in short supply and come at high cost."
Webinar
Sep
25
CMS Briefing: A Live Look at What’s Next in AI-Driven Platforms
Learn how leading organizations are using AI‑driven tools to publish faster, personalize smarter and stay secure.
Register
Webinar
On demand
Ready or Not: How Data-First Organizations Are Unlocking Agentforce Potential
Learn how to cut through the noise, activate Agentforce and build a Salesforce AI strategy that actually delivers.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
Webinar
On demand
From Hype to High-Impact CX Strategies That Actually Scale
Turn buzzworthy AI and outsourcing trends into measurable CX wins with fresh 2025 data.
Watch Now
Webinar
On demand
Insights to Action Rethinking the Contact Center for Real Business Impact
Join our exclusive webinar to hear CX executives share their innovative strategies for transforming service delivery.
Watch Now
Webinar
On demand
Accelerating Healthcare Ops with AI
Sign up to see how your peers are putting AI to work — and making it stick.
Watch Now
Enterprise Use Cases
The enterprise use-cases are already clear, said Cyphere managing consultant Harman Singh. Watsonx offers clients and partners the opportunity to specialize and deploy models for their specific enterprise use cases, meaning businesses can tailor its AI capabilities to suit their specific needs, leading to more accurate and effective results.
He noted the impact of the 150 enterprises in the beta and tech preview programs in helping shape the platform, which shows Watsonx has been assessed and validated in real world applications, ensuring it is reliable and robust.
“The diverse range of industries represented by these users are from telcos to banking, demonstrates the versatility of Watsonx. It can be utilized in different sectors, enabling organizations across industries to benefit from its AI capabilities,” Singh said.
Overall, the significance of this release lies in the opportunities it provides for customization, the validation it has received from real users, and its versatility across industries, he continued. "With Watsonx, businesses can now harness the power of AI in a way that aligns with their unique requirements, ultimately driving better outcomes and enhancing their overall operations."
Enterprise-Ready AI
IBM's Watson was one of the first big names in AI for the public, said Dave Jenkins, VP of digital technologies at Iterate.ai. So while the spotlight recently shifted to ChatGPT and other ‘consumer-level' AI engines, enterprises are demanding 'enterprise-level' AI engines.
“If this Watsonx evolution is geared more toward enterprise deployment, I think we could certainly see some very powerful capabilities come to market,” he said. "Companies are increasingly demanding private large language models (LLMs), and AI engines that are specifically tuned to their own core reference data."
It will be important to draw the line between public baseline general shared knowledge and enterprise-client specific knowledge, he concluded. That distinction will be important in the end.