There is one notable corner of the tech world that has not been touched by the artificial intelligence euphoria sweeping through the stock market.
If generative AI really does represent the next great sales opportunity for the tech industry, then software companies ought to be among the biggest winners. After all, most AI is likely to show up as enhanced features in the business software that companies rely on in their daily operations.
However, the BVP Nasdaq index of cloud software companies is down nearly 10 per cent this year, while the Nasdaq Composite is up more than 20 per cent. It has also halved from its pandemic-era peak. The slump points to an industry at a crossroads. A long, secular growth phase driven by the rise of the cloud looks like it is entering a new and more mature state, while the next (the spread of generative AI in business) has barely begun.
At times like this, Wall Street faces complex questions. If the cloud business really is maturing, the focus of investors needs to shift more quickly from growth to value. Tech companies that recently reported disappointing results, such as Salesforce, MongoDB and Workday, have tried to pass the lull off as the result of extended economic weakness. But the longer it goes on, the harder this argument is to sustain. Salesforce’s revenues doubled in the past four years to $36bn: at that scale, the slower 10 per cent growth it has projected for next year begins to look more like the norm.
At the same time, investors have to handicap which companies will catch the next wave of growth and which will fail to adapt and be left in the dust.
According to the companies themselves, the lack of an impact on their sales from AI is just a timing issue. Salesforce chief executive Marc Benioff, for instance, points to the challenge of training large armies of salespeople to handle the harder, more complex task of selling AI services. Customers are grappling with a wide set of questions, seeking to understand how the new AI models work and how their workers should interact with them. They also need to consider how to redesign their work processes to make best use of the technology, as well as deal with new threats to the security of their data.
Even if sales are still negligible, the software companies report huge interest from customers in piloting their new AI services. This may mean the AI dividend has just been delayed.
Yet the disruptive threats from AI suggest things will not be so straightforward. One is the upheaval to the cloud companies’ business model. Most rely on charging per-seat subscriptions, meaning their revenue goes up in line with the number of workers using their services. If generative AI works as promised and makes workers far more productive, customers should be able to do more with fewer staff.
The result has been a pivot towards consumption-based pricing, or charging based on how much the new services are actually used. Tying fees to usage has the added advantage of offsetting some of the higher cost of delivering generative AI. But unless this leads to real and demonstrable business benefits, the software companies could face a backlash when customers see their bills soar.
The software groups also have tech history to contend with. In the past, new tech eras — such as the rise of client-server computing in the 1990s and cloud computing the following decade — have brought new waves of start-up software companies to the fore. New companies, their products and business models designed from the ground up to fit a new computing paradigm start with a big advantage.
The first wave of these “AI native” software companies has often looked like little more than “wrappers” around the large language models, adding only a veneer of industry-specific expertise as they offer businesses ways to adopt generative AI. But they are all working hard to gain a foothold from where they can start to build out more compelling services.
According to Salesforce’s Benioff, the incumbents will be hard to unseat. Companies such as his have become the repositories of their customers’ most important data, he says, giving them a big advantage when it comes to training the AI models that businesses will find truly useful.
That will only count if today’s cloud companies can adapt their own products and processes to the new technology fast enough. For now, Wall Street is suspending judgment.