Firms Strain to Price Machine Learning

"More than two years after the public debut of ChatGPT, software companies still haven't found a compelling way of charging for AI tools, chief information officers say. Now they are trying new strategies.

In the past couple of years, vendors have typically charged a monthly fee per user for AI features, assistants and copilots, much like they price other software as a service. But AI's high compute demands meant they needed to charge eyebrow-raising prices to cover the cost of delivering the service.

For example, some chief information officers balked at paying $30 per user per month to add Microsoft's AI Copilot to its 365 productivity suite, a 60% premium to the top level of 365 without AI.

"A year ago everything was way overpriced," said Greg Meyers, chief digital and technology officer of Bristol-Myers Squibb. "Most companies overestimated how much more we would be willing to pay for an AI feature."

The emergence of models like DeepSeek's R1, which the Chinese company said it trained for a fraction of the cost of leading U.S. models, could help drive AI costs down over time.

But meanwhile, CIOs remain in a tough spot.

"We're in a place where prices are high and simultaneously companies are trying to understand how to drive value out of it," said United Airlines Chief Information Officer Jason Birnbaum.

When it comes to general-purpose tools like Copilot that are charged on a per-seat basis, Birnbaum added, "We're not really ready to deploy it on a broad basis."

Now vendors are making changes to how they price in an attempt to gain more users and more adoption.

Alphabet's Google in January said its Business Standard plan would shift from charging $12 per person per month for its Workspace productivity suite, plus another $20 for access to its Gemini AI business tools, to a $14 package with Gemini AI features baked into Workspace.

And Microsoft introduced consumption-based pricing with its new Microsoft 365 Copilot Chat that gives users access to use AI agents. Depending on the interaction, customers might pay a few cents for each "use."

To be sure, Microsoft said it is seeing accelerated customer adoption of the $30 Copilot offering. Chief Executive Satya Nadella said on Microsoft's earnings call last week that customers who purchased Copilot during its first quarter of availability "expanded their seats collectively by more than 10X over the past 18 months."

But the goal with the new Copilot Chat is also to lower the barrier to entry for new enterprises using Copilot and build a broader user base that will ultimately use the $30-per-month version, said Jared Spataro, chief marketing officer of AI at Work for Microsoft.

"A per user per month charge can sometimes be difficult for them if they're trying to go to broad scale because they're just not sure how to value something," Spataro said.

Kathy Kay, CIO of Principal Financial, said she plans to test the new Copilot Chat tool to determine its cost-effectiveness.

"I do think this is an evolution of companies listening to their customers and understanding what they need and making it much more cost effective," Kay said.

But the vendors face another threat: Enterprises can access the same underlying models they do to build similar tools of their own. Kay said she built a tool that was able to replicate some of the capabilities of Copilot at a much lower cost.

Amazon Web Services is in part betting on that strategy. Its Bedrock platform allows users to access models from companies like Anthropic, Meta Platforms and Mistral AI with either a no-commitment, pay-as-you-go pricing model starting at less than one cent per interaction or a time-based term commitment, starting at $25 per hour of commitment to use the Bedrock service. Amazon.com also provides its work assistant, Amazon Q, for $3 to $20 per user per month, depending on the tier.

Software companies also are facing pressure to adapt their pricing to account for the fact that the actual cost of using the underlying models is going down. As that happens, CIOs don't want to feel like their vendors are simply taking a bigger share of the profits.

"If they aren't fair and equitable in how they price those tools and transactions, they're actually going to incent me to build my own capability over buying theirs," said Nationwide Chief Technology Officer Jim Fowler. "And so my biggest concern is in this rush to AI, that they price themselves out of the enterprise."

Vendors and enterprises alike are still working to figure things out, he said. "It's still the wild west."

Salesforce says it is targeting more flexibility when it comes to their pricing options. Last September, the company rolled out a pricing plan that allowed enterprises to toggle their spend minimums between per-month licenses for human employees and consumption-based agents.

A lot of customers are still trying to make sure they have the right value equation, said Bill Patterson, executive vice president of corporate strategy at Salesforce, and for some of the AI investments companies have made over the past two years, the jury is still out.

Meanwhile, vendors continue facing the dilemma of making tools cheap enough that people will buy them but expensive enough so they aren't losing money in compute costs if people use it too much -- a balance that is hard to navigate with tools that are so new.

Earlier this year, OpenAI CEO Sam Altman posted on X that the $200-a-month ChatGPT Pro plan was losing money because people were using it more than anticipated.

(The ChatGPT Enterprise plan is separate and typically comes in at about $30 to $45 per seat, OpenAI said).

Going forward, CIOs anticipate more changes and experimentation with different pricing strategies from their vendors.

"We're in such an interesting and fluid time, it's hard to say which variant is going to win," said Don Vu, chief data and analytics officer at New York Life." [1]

1. Firms Strain to Price Machine Learning. Bousquette, Isabelle.  Wall Street Journal, Eastern edition; New York, N.Y.. 03 Feb 2025: B1.