Salesforce Unveils AIforce and Headless 360 at Dreamforce as Nscale Files for IPO

Salesforce CEO Marc Benioff bantered with Nvidia CEO Jensen Huang and led the announcement, the report said. The event reflected Salesforce's ambition to create what it calls an Enterprise AI Harness, which combines data, business knowledge, workflows and control into a common, composable architecture for agents. Salesforce has fully committed to agents, a strategy that appears to be paying off so far, according to SiliconANGLE.

On theCUBE Pod, John Furrier, executive analyst at theCUBE Research, said the headless offering shows AI agents are not another feature cycle. Agents are not a feature of the enterprise, he said; they are forcing a complete reconstruction of the enterprise stack from applications and end interfaces down through security, infrastructure and capital allocation.

Dave Vellante, chief analyst at theCUBE Research, said Headless becomes the system of intelligence, where data is harmonized and process knowledge is injected. AIforce allows any client surface to create the interface, he said. The technology is not mature, and Agentforce is still in its early days, but there is clear business value.

Salesforce has fully committed to agents, perhaps to avoid the potential SaaSpocalypse that is still spooking investors, according to the podcast. Much of its new stack is formed from composable agents, which give users greater flexibility and security in theory. Furrier said Salesforce's advantage might be the remaining system of data, the workflow, the identity data and the business context underneath it all. That is why Slack becomes a potential agentic front door, he said, because unstructured data works well with graphs and with the infrastructure AI likes.

The on-premises renaissance has continued and become part of the bedrock of AI revenue. Vellante noted that Salesforce is still figuring out its pricing. Research from Qualitate suggests that 75% of respondents who had modeled or experienced the financial impact of headless access expect their Salesforce spending to increase, not decline. Vellante said Claude is creating the interface and Slack is creating the interface. Virtually any harness will be able to create the interface, he said, which will drive token demand. That token demand will come from core infrastructure in the data center, and much of it will be on-prem, he added.

Breaking news during the podcast was Nscale's IPO filing. The Nvidia-backed company builds AI-optimized data centers and has yet to become profitable. It is betting on continued growth in demand for its data center capacity. Furrier said Nscale is an example of the playbook CoreWeave built: a fully integrated business that went full stack and bought Anyscale. He described a huge cloud provider opportunity for AI infrastructure serving enterprises, with Nebius and CoreWeave facing a new competitor. The IPO is expected to be pretty smashing, he said.

Furrier also analyzed the development of agents in the enterprise. AI is starting to make more autonomous decisions based on enterprise operational data, and with Salesforce's new headless offering, agents could fundamentally reshape how businesses operate, he said. He asked what happens when agents start to learn how enterprises actually work over time, creating a new training and inference dataset. He pointed to Spirit Airlines' bankruptcy and a bid for 20 years of data, with buyers seeking it for training purposes. That raises the commoditization of the system of record and the question of when it becomes disposable.