Every generation of software moves interfaces closer to where people think and work. Terminals gave way to GUIs. GUIs gave way to browsers. And now, the interface is language itself. Conversation has become the most natural way to build, explore, and decide. At the center of this shift is a new pattern: the AI agent. Today, software doesn’t have to wait for clicks or configuration, but understands user intent, reason about it, and takes action. The question for enterprises isn’t if they’ll adopt agents, but where those agents will live. Our answer: where work already happens. That’s why Vercel and Salesforce are partnering to help teams build, ship, and scale AI agents across the Salesforce ecosystem, starting with Slack. Together, we’re bringing the intelligence and flexibility of the Vercel AI Cloud to the places teams collaborate every day.

When OpenAI announced the Apps SDK with Model Context Protocol (MCP) support, it opened the door to embedding web applications directly into ChatGPT. But there's a significant difference between serving static HTML in an iframe and running a full Next.js application with client-side navigation, React Server Components, and dynamic routing. This is the story of how we bridged that gap. We created a Next.js app that runs natively inside ChatGPT's triple-iframe architecture, complete with navigation and all the modern features you'd expect from a Next.js application.

As AI reshapes how software is built and deployed, the surface area for attacks is growing rapidly. Developers are shipping faster than ever, and we’re seeing new code paths, new threat models, and new vulnerabilities. That’s why I’m excited to share that Talha Tariq is joining Vercel as our CTO of Security. Talha brings deep expertise in security at scale, having served as CISO & CIO at HashiCorp for seven years before becoming CTO (Security) at IBM following its acquisition. There, he oversaw security across all IBM divisions including software, AI, and post-quantum cryptography.

For teams on Vercel, Black Friday is just another Friday. The scale changes, but your storefronts and apps stay fast, reliable, and ready for spikes in traffic. Many of the optimizations required for peak traffic are already built into the platform. Rendering happens at the edge, caching works automatically, and protection layers are on by default. What’s left for teams is refinement: confirming observability is set up, tightening security rules, and reviewing the dashboards that matter most. Last year, Vercel created a live Black Friday Cyber Monday dashboard that showcased our scale in real-time, showing the spikes. Overall, from Friday to Thursday, Vercel served 86,702,974,965 requests across its network, reaching a peak of 1,937,097 requests per second. Helly Hansen, a major technical apparel brand, entered the weekend with this confidence. Before the event, they moved from client-heavy rendering to Vercel’s CDN and saw:

Independent developer Theo Browne recently published comprehensive benchmarks comparing server-side rendering performance between Fluid compute and Cloudflare Workers. The tests measured 100 iterations across Next.js, React, SvelteKit, and other frameworks. The results showed that for compute-bound tasks, Fluid compute performed 1.2 to 5 times faster than Cloudflare Workers, with more consistent response times.

Today, Vercel announced an important milestone: a Series F funding round valuing our company at $9.3 billion. The $300M investment is co-led by longtime partners at Accel and new investors at GIC, alongside other incredible supporters. We're also launching a ~$300M tender offer for certain early investors, employees, and former employees. To all the customers, investors, and Vercelians who have been on this journey with us: thank you.

Claude Sonnet 4.5 is now available on Vercel AI Gateway with full support in AI SDK. We’ve been testing the model in v0, across our Next.js build pipelines, and inside our new Coding Agent Platform template. The model shows improvements in design sensibility and code quality, with measurable gains when building and linting Next.js applications. Claude Sonnet 4.5 builds on Anthropic's strengths in reasoning and coding. When paired with the Vercel AI Cloud, it powers a new class of developer workflows where AI can plan, execute, and ship changes safely inside your repositories.

When you deploy a Next.js app with Incremental Static Regeneration (ISR), pages get regenerated on-demand after their cache expires. ISR lets you get the performance benefits of static generation while keeping your content fresh. But there's a problem. When many users request the same ISR route at once and the cache is expired, each request can trigger its own function invocation. This is called a "cache stampede." It wastes compute, overloads your backend, and can cause downtime. The Vercel CDN now prevents this with request collapsing. When multiple requests hit the same uncached path, only one request per region invokes a function. The rest wait and get the cached response. Vercel automatically infers cacheability for each request through framework-defined infrastructure, configuring our globally distributed router. No manual configuration needed.

Your traffic is spiking and you spot suspicious bot activity in your logs. You deploy BotID expecting to find malicious scrapers, but the results show verified Google bots. Normal crawlers doing their job. But then you notice what they're actually searching for on your site. Queries that have nothing to do with your business. What do you do? This exact scenario recently played out at one of the largest financial institutions in the world. What they discovered was a years-old SEO attack still generating suspicious traffic patterns.

Recently, we shipped an optimization to our global routing service that reduced its memory usage by 15%, improved time-to-first-byte (TTFB) from the 75th percentile and above by 10%, and significantly improved routing speeds for websites with many static paths. A small number of websites, with hundreds of thousands of static paths, were creating a bottleneck that slowed down our entire routing service. By replacing a slow JSON parsing operation with a Bloom filter, we brought path lookup latency down to nearly zero and improved performance for everyone.

In February 2025, Andrej Karpathy introduced the term vibe coding: a new way of coding with AI, “[where] you fully give in to the vibes, embrace exponentials, and forget that the code even exists.” Just months later, vibe coding is completely reshaping how developers and non-developers work. Over 90% of U.S. developers use AI coding tools, adoption is accelerating for other roles, and English has become the fastest growing programming language in the world. We explore this shift in detail in our new State of Vibe Coding. Here are a few of the key takeaways.

Cold starts have long been the Achilles’ heel of traditional serverless. It’s not just the delay itself, but when the delay happens. Cold starts happen when someone new discovers your app, when traffic is just starting to pick up, or during those critical first interactions that shape whether people stick around or convert. Traditional serverless platforms shut down inactive instances after a few minutes to save costs. But then when traffic returns, users are met with slow load times while new instances spin up. This is where developers would normally have to make a choice. Save money at the expense of unpredictable performance, or pay for always-on servers that increase costs and slow down scalability. But what if you didn't have to choose? That’s why we built a better way. Powered by Fluid compute, Vercel delivers zero cold starts for 99.37% of all requests. Fewer than one request in a hundred will ever experience a cold start. If they do happen, they are faster and shorter-lived than on a traditional serverless platform. Through a combination of platform-level optimizations, we've made cold starts a solved problem in practice. What follows is how that’s possible and why it works at every scale.