Vercel said on 15 October 2026 that Meta Superintelligence Labs runs on Vercel through its Bring Your Own Cloud option on AWS, in an enterprise post published as the company held its Vercel Ship conference in San Francisco. The same day it made Vercel Drives, persistent storage for AI agents, generally available.
In the last three months, token usage by enterprise companies on its AI Gateway has quadrupled, Vercel says. The Stack’s headline on its interview with CEO Guillermo Rauch, published the same day, says 60% of activity on Vercel is now driven by coding agents.
Business Pill · AI AGENTS
A one-minute explainer of AI agents: models that move from answering to doing. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, Vercel is repositioning from hosting websites toward running agents and backends: on the day it made Drives generally available it also says enterprise token usage on its AI Gateway quadrupled in three months and names Meta Superintelligence Labs as a customer on Bring Your Own Cloud.
Vercel Drives is generally available for Vercel Sandbox on all plans, the company says. Drives default to 1 TiB, or 1 GiB on the Hobby plan, with up to 16 TiB available.
Vercel Daemons, which runs containers as continuously running background workers, is in public beta, according to Vercel’s changelog. Python on Vercel is now generally available, and the company says 150,000 teams already run Python backends on it, including engineers at OpenAI, Polymarket and Exa.
Vercel says that since container support entered public beta, 5,000 teams have shipped 150,000 container deployments. On the enterprise side, Bring Your Own Cloud on AWS is generally available, and Enterprise-Managed Authorization through Vercel Connect is launching in public beta, giving agents access to connected services on behalf of employees through a company’s identity provider.
Vercel says token usage by enterprise companies on its AI Gateway has quadrupled in the last three months, that the average enterprise customer routes across 17 models, that 150,000 teams run Python backends on Vercel and that its own Vertex agent handles 91% of its customer support cases (Vercel, 15 October 2026).
Vercel’s own account announced on 15 October 2026 that Vercel Drives is generally available, saying users can now run stateful agents on Vercel, with a link to the launch post this piece draws on.
Why Agents Need a Drive
“We built Vercel Drives to keep an agent’s files independent of its execution environment,” Vercel writes. Agents can mount the same drive in a new sandbox configured for the next task and continue where they left off, the post says.
Other agents can mount read-only copies of a drive to analyze its data without rebuilding it, and agents can fork a drive to try different approaches in parallel. “Forks don’t merge back into the original drive,” the post notes.
Vercel says Meta Superintelligence Labs runs on Vercel via Bring Your Own Cloud, inside its own AWS environment. “Leaders at Meta said that in their first week on Vercel, they shipped a month worth of work,” the post says.
The average enterprise company using its AI Gateway routes across 17 different models, according to Vercel, which says the gateway charges no markup on cost. Its customer examples: Mercor processed 63 billion tokens in its first 10 days on AI Gateway, and at Klaviyo 512 employees deployed 356 apps in their first two weeks on Vercel.
Vercel also describes its own customer support agent, Vertex: “Today, Vertex handles 91% of our customer support cases autonomously, processing thousands of tickets a week.” The spending race to put engineers alongside customers deploying agents is one we covered in our piece on forward deployed engineers.
What Rauch Told The Stack
“Every organization is going to have to start exposing their data and their capabilities through APIs, through MCPs, and through CLIs,” Rauch told The Stack. “The next chapter of Vercel, the workload shape is going to be a backend, it’s going to be an API server,” he said, adding: “It’s going to be a lot more long-running.”
The Stack says Vercel has raised $863 million over the last ten years and calls it a $9.3 billion company. Its headline puts 60% of activity on Vercel as now driven by coding agents.
The Structural Read
As we read it, Drives addresses the part of an agent that outlives a single run: Vercel says it built Drives to keep an agent’s files independent of its execution environment, so a new sandbox can pick up where the last one stopped.
As we read it, the enterprise pitch is governance rather than raw compute: Vercel describes Enterprise Apps and Agents as a central control plane for governing every builder, app, agent, token and dollar, with Bring Your Own Cloud keeping compute inside a customer’s AWS account.
As we read it, Rauch frames the shift in workload terms: he told The Stack that the workload shape is going to be a backend and an API server, and a lot more long-running.
Guillermo Rauch, Vercel CEO, to The Stack
“Every organization is going to have to start exposing their data and their capabilities through APIs, through MCPs, and through CLIs”
Three Implications
AGENT BUILDERS Vercel says Drives is generally available for Vercel Sandbox on all plans, with drives defaulting to 1 TiB and up to 16 TiB available.
ENTERPRISE BUYERS Vercel says Bring Your Own Cloud on AWS is generally available and that Meta Superintelligence Labs uses it to run inside its own AWS environment.
MODEL PROVIDERS Vercel says the average enterprise company on its AI Gateway routes across 17 different models, with no markup on cost.
The framework puts it this way: “Enterprise vendors don’t sell software anymoreโthey sell irreversibility.”
As we read it, Vercel’s launches push in that direction: drives that hold agents’ files across runs, a control plane over every app, agent, token and dollar, and compute that runs inside a customer’s own AWS account. The framework does not discuss Vercel.
What Vercel shipped on 15 October 2026, per its blog and changelog: Drives, Python and Bring Your Own Cloud on AWS are generally available; Daemons and Enterprise-Managed Authorization are in public beta. Vercel says Meta Superintelligence Labs runs on Vercel via BYOC.
What Is Not Established
The rest of The Stack’s interview is for members only and we did not read it, so we cannot say what the 60% figure measures, whether deployments, requests or another metric.
The Vercel posts we read give no prices for Drives or Daemons beyond plan availability. The Meta remark is relayed by Vercel, and the Vertex, Mercor and Klaviyo figures are Vercel’s own.
Business Engineer Framework
Enterprise AI: From Software to Substrate
A Business Engineer framework on how enterprisesoftware turns from a tool you buy into infrastructure you are woven into.
Vercel made persistent Drives for AI agents generally available on 15 October 2026 and says token usage by enterprise companies on its AI Gateway has quadrupled in three months.
It names Meta Superintelligence Labs as a customer running on Vercel through Bring Your Own Cloud on AWS.
A note on sourcing. We read Vercel’s three 15 October 2026 posts (“Vercel for Enterprise”, “Introducing Vercel Drives” and “Backends on Vercel”) in full, plus its changelog listing; every product detail, customer example and usage figure comes from them and is Vercel’s own. The Rauch quotes, the funding and valuation figures and the 60% headline come from The Stack’s interview; we read its public introduction, and the rest is for members only. The Perplexity and forward-deployed-engineer context comes from our own earlier reporting.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
