Artificial intelligence is incredible. ChatGPT, a chatbot from OpenAI powered by a large language model, can do amazing things. Companies big and small are racing to incorporate AI into their own products, afraid of being left behind. It's a gold rush and an arms race all in one.
But there's a problem: AI is expensive. Training a large language model can require thousands of high-powered data center GPUs that are in short supply. And because these models require mammoth amounts of memory and computational horsepower to be used once trained, they generally must run in the cloud, not on users' devices.
Price hikes will test AI's appeal
The AI features being layered on top of popular software products won't come for free. While the potential for AI software in the long run is enormous -- some estimates put it at a $14 trillion market by 2030 -- software providers must first convince customers to spend more. In the current economic environment, that might be a tough sell.
Salesforce (CRM 2.46%), a leading provider of cloud-based customer relationship management software, has been busy adding a slew of generative AI features to its suite of software products. Customers are now going to have to pay for those features. Starting in August, Salesforce is boosting prices for the first time in 7 years. On average, list prices will go up by about 9%.
Salesforce's products are sticky, so most customers are likely to accept the price increase without much fuss for now. But how much value do these AI tools really provide? They're being forced on Salesforce's customers, who will have no choice but to pay for them. Any customer who doesn't find them useful may start to think about jumping ship.
Microsoft (MSFT 1.26%) is another example of a company that's heavily invested in AI and now must get its customers to pay for it. Microsoft invested $10 billion into OpenAI earlier this year, and the company has been working to integrate OpenAI's technology into its own products. Microsoft 365 Copilot will be the first real test of whether customers are willing to pay for these AI capabilities.
Microsoft 365 Copilot brings AI to the company's Office suite, enabling users to generate text, create presentations in PowerPoint, and visualize data in Excel with simple prompts. The tool can also summarize long emails, automate tasks using Microsoft's Power Platform, and do a variety of other productivity-boosting tasks.
All that power will come at a high cost. For commercial customers, 365 Copilot will be priced at $30 per user per month on top of the existing pricing for Microsoft 365 plans. The Microsoft 365 Business Premium plan, which includes all the company's Office software as well as advanced features, is priced at $22 per user per month. Adding 365 Copilot will more than double the total per-user price.
Potentially a tough sell
How many businesses are going to be clamoring to more than double their spending on Microsoft Office software? The AI features being introduced are impressive, but it's hard to say whether they'll boost productivity enough to justify the sky-high price tag.
That $14 trillion estimate for the size of the AI software market by the end of the decade assumes that businesses will be willing and able to vastly grow their spending on software. It's not obvious to me that will be the case. If these AI add-ons fail to provide a meaningful productivity boost, businesses won't be willing to pay anything for them. Worse, they might actually decrease productivity. Making it easier to create PowerPoint presentations doesn't necessarily sound like a good thing.
The long-term growth of the AI industry ultimately depends on businesses being willing to pay for AI-powered products. High prices for those products could slow adoption and lead to slower growth than expected over the next decade.