International Business Machines (IBM -0.94%) has been working on artificial intelligence (AI) technology for decades, but the company is only now finding its footing and turning AI into a viable business. IBM's main focus is enabling large enterprises to leverage generative AI to boost productivity and unlock efficiencies. That's a powerful pitch, given that large investments in AI often have unclear and uncertain returns on investment.
Thinking small
IBM has developed its own AI models tailored for enterprise use. The latest iteration, the Granite 3.0 family, goes up to 8 billion parameters. These are small models relative to the most powerful generative AI models, which are estimated to have hundreds of billions of parameters. In general, the more parameters a model has, the more capable the model is.
While IBM's Granite 3.0 models aren't going to outclass the best models available, they're designed to be fine-tuned by enterprise customers with proprietary data. A small model that's been fine-tuned can be dramatically more capable in some areas than the base model. As part of its suite of AI tools, IBM has developed an open-source project called InstructLab that allows users to iteratively and collaboratively fine-tune an AI model without starting from scratch each time.
A small AI model is much cheaper to run than a large, best-in-class model. IBM estimates its small Granite 3.0 models that have been fine-tuned can produce task-specific performance that's close to the top-tier AI models, while costing anywhere from 3 times to 23 times less to run. That's a huge win for enterprises.
Taking things further, IBM's Granite 3.0 family includes even smaller models that are designed for low-latency applications and can be run directly on a CPU rather than a pricey AI accelerator. For enterprises looking to run AI workloads on their own hardware, these small models don't require big outlays for data center GPUs.
AI momentum
IBM is scheduled to report its third-quarter results on Oct. 23. The company will likely update investors on the progress of its AI business, which has been booming in recent quarters. At the end of the second quarter, IBM had booked a cumulative $2 billion worth of generative AI-related business. That number stood at "greater than $1 billion" three months earlier.
Key to IBM's AI growth is its consulting business. Around two-thirds of the company's AI-related bookings have come from consulting signings, a strong sign that enterprises are looking for AI expertise, guidance, and implementation services in addition to software and models. Those consulting contracts can then drum up business for IBM's various software businesses, including its watsonx AI platform.
AI is still a small part of IBM's overall business, but it's growing rapidly each quarter. The company expects to generate more than $12 billion in free cash flow this year, and AI should help boost that number further in 2025 and beyond. Even if the state of the global economy takes a turn for the worse, IBM's focus on enabling customers to use AI technology to save money and boost productivity should continue to resonate with its customer base.
With IBM trading for less than 18 times free cash flow guidance, the stock still looks like a decent value, given the company's enterprise AI potential.
