Salesforce Agentforce AI growth has reached more than $1.5 billion in annualised revenue, a 240% year-on-year increase, according to MarketScale, as fears that artificial intelligence would hollow out the software-as-a-service sector continue to unravel.
Salesforce, Booking Holdings and IBM are among the companies that have absorbed the heaviest AI-driven sell-offs yet are emerging as potential long-term beneficiaries, not casualties, of the technology.
Salesforce Agentforce AI growth and the Claudeforce deal
Salesforce’s stock has fallen roughly 20% this year and around 40% from its peak, priced by bears for an obsolescence that has not materialised.
The pessimistic case held that large language model companies such as OpenAI and Anthropic would manage customer relationships end-to-end, reducing Salesforce to a passive database. Critics at Wells Fargo rejected that reading, declaring that ‘lower cost of intelligence increases value of incumbent data.’
Salesforce processed over 216 trillion customer records this year and ingested 104 trillion records in a single quarter alone, double the figure from the prior quarter. That volume of structured, trusted customer data is what AI agents require to function.
The Salesforce Agentforce AI growth story accelerated further with the announcement of Claudeforce, a partnership with Anthropic. According to Salesforce Investor Relations, the Claudeforce Plugin includes 37 prebuilt sales skills that enable sellers and agents to reason over live revenue context, automate pipeline updates, and take governed action within Salesforce.
Agentforce has closed well over 30,000 deals. Slack, acquired by Salesforce for $25 billion, is also playing a growing role: Slackbot users grew over 150% quarter on quarter, and when Salesforce opened Slack to outside AI agents, one million users connected within a month.
Chief executive Marc Benioff spent $25 billion buying back Salesforce stock in a single quarter, described as the largest repurchase in the company’s history and roughly a fifth of its market capitalisation.
Booking Holdings and IBM: infrastructure over disruption
Booking Holdings faced a different version of the same bearish thesis: that chatbots would replace travel search platforms. The stock has since recovered to near all-time highs.
Booking.com acts as merchant of record on roughly three-quarters of its bookings, a share up four points in the past year, settling more than 100 payment methods across 50 currencies. Around 90% of its room nights come from independent properties and smaller hotels that could not independently run global payment processing or dispute resolution.
Google’s own leadership stated the company has ‘no intention of becoming an OTA,’ and OpenAI retreated from in-chat checkout after a widely criticised rollout. On its August earnings call, Booking disclosed that traffic sourced from large language models remains well below 1% of room nights, with direct traffic holding steady in the mid-60% range.
IBM’s situation is equally misread. Its stock fell 25% in a single day, the worst in the company’s history, after several large clients redirected budgets towards memory during a supply crunch. A third of those deals subsequently closed within weeks.
AI now accounts for half of all new IBM consulting signings and forms one of the largest components of its backlog, at margins higher than traditional consulting because IBM can bill on outcomes rather than hours. Red Hat, which allows AI agents to run across any cloud or platform, grew 11%. IBM’s AI business has more than doubled over the past year.
The common thread across all three is ownership of infrastructure that AI agents cannot operate without: Salesforce holds the customer data, Booking Holdings holds the payment and settlement layer, and IBM holds the underlying platforms on which agents run.
The Claudeforce Plugin’s 37 prebuilt skills, detailed by Salesforce Investor Relations, represent the clearest example yet of frontier AI models choosing to integrate with incumbent data platforms rather than displace them, a structural shift that the original SaaSpocalypse thesis did not anticipate.
