Convex Raises $57M Series B to Scale Its Agentic Backend Platform

Convex announced a $57M Series B led by Insight Partners on August 4, 2026. Etna Labs, Spark Capital, a16z, and Justin Kan participated in the financing, which arrives less than a year after Convex announced a separate $24M round.

The San Francisco company is building a reactive backend platform for software created by developers and coding agents. Convex combines a database, server functions, workflows, real-time sync, search, file storage, and sandboxed Components in one TypeScript system designed to make correctness and scale less dependent on humans inspecting every implementation detail.

That distinction is why the round matters. AI has made code generation cheaper and faster, but production software still needs durable transactions, safe boundaries, predictable synchronization, and infrastructure that does not collapse when the demo becomes a business. Convex is raising capital to make those guarantees part of the platform rather than another assignment on an engineering team's growing list.

What Happened

According to the official Series B announcement, Insight Partners led Convex's $57M round, with Etna Labs, Spark Capital, a16z, and Justin Kan joining. Convex did not disclose a valuation, board change, or investor check sizes. The company said the capital will help it build the platform to the standard and scale customers are requesting while hiring across engineering, engineering leadership, product, marketing, and go-to-market.

The financing follows a $24M round announced in November 2025, led by a16z and co-led by Spark Capital, and a $26M Series A announced in April 2022, led by a16z. Those three disclosed rounds sum to at least $107M, excluding an earlier seed whose amount was not stated in the reviewed primary sources. The total is a derived calculation, not a company-published cumulative funding figure.

Why Agent-Written Software Changes the Backend

The old application stack assumed that people would understand most of the code they shipped. That assumption was always generous, but it at least shaped the rituals: developers selected infrastructure, debated database tradeoffs, reviewed implementation details, and carried architectural context from one pull request to the next. Coding agents compress that cycle and make the human review budget the scarce resource.

Convex's argument is that infrastructure must absorb more of the judgment. The platform advertises ACID transactions so concurrent writes do not corrupt data, end-to-end TypeScript so invalid fields fail before production, automatic caching and scaling, and real-time synchronization without a separate layer of hand-built glue. These are familiar backend concerns, but agentic development changes their volume. A system producing more code also produces more opportunities for hidden assumptions to become expensive.

Convex Components push the idea further. Each Component is a sandboxed backend with its own schema, functions, data, and explicit API. That boundary gives an agent a smaller unit it can reason about completely while limiting its ability to break invariants elsewhere in the application. The market pitch is not that agents become perfect. It is that the platform should assume they are not.

The Evidence Behind the Raise

Convex says millions of its instances are running in production and dozens of companies are already using its Enterprise plans. When the company launched Convex for Enterprise in April 2026, it reported nearly 10,000 paying teams and named Conduit, doxy.me, Orchid, Reducto, and Town among Enterprise customers. Those figures are company-reported, but they establish a commercial base beyond experimental weekend projects.

The progression since the prior round is also visible in the product. Convex has launched European hosting for teams that need EU data residency and added Enterprise options for dedicated deployments, SSO, audit logging, support agreements, and procurement requirements. The current financing is therefore not a blank-check bet on a prototype. It is a bet that demand from larger customers will keep turning a developer-friendly backend into durable infrastructure.

The numbers still leave important blanks. Convex has not disclosed revenue, ARR, market share, a current audited growth rate, or the precise split between hiring and infrastructure investment. Funding stories get unreliable when every missing figure is replaced with confidence. Here, the supported signal is simpler: more production usage, more paying teams, and a larger capital base are meeting at the point where software creation is changing fastest.

The Founders Built the Pain Before Selling the Cure

Jamie Turner, Co-Founder and CEO, and James Cowling, Co-Founder and CTO built their careers around distributed systems before Convex existed. Turner led storage, database, and business-platform engineering at Dropbox after serving as Head of Engineering at Bump. Cowling was a Senior Principal Engineer at Dropbox and helped build its multi-exabyte storage system after doctoral work at MIT in distributed transaction processing and consensus.

Sujay Jayakar was also part of the original three-person founding team named by a16z in 2022, though he is not listed on Convex's current team page and should not be presented as a current executive. That distinction matters because accurate funding coverage should preserve company history without turning an old title into a current role. The active operating founder roles are held by Turner and Cowling.

This background gives the pitch unusual coherence. Convex is not trying to bolt an AI label onto a generic database service. Its founders spent years operating storage and synchronization systems at enormous scale, then built a platform that moves those hard-won guarantees closer to ordinary developers and the agents working beside them.

Why Insight Partners Fits the Next Phase

Insight Partners describes itself as a global software investor, and the Series B introduces a lead investor associated with helping software companies scale. Returning participation from a16z and Spark Capital matters for a different reason: both firms had already watched Convex move from its infrastructure thesis through product adoption and into an Enterprise offering.

The financing also lands during a shift in developer tooling economics. Model providers are competing to generate better code, while application teams are discovering that cheap code can make weak architecture more expensive. Backend platforms that enforce transactions, type safety, isolation, and synchronization are positioned to collect value from that tension, provided they remain simple enough for agents and humans to use without recreating the complexity they promised to remove.

Convex is not alone in chasing the agentic application stack, and capital does not settle the architecture. The harder test is whether developers keep choosing a proprietary abstraction when agents make switching easier and when familiar databases retain deep ecosystems. Convex's answer is to make the abstraction earn its keep through guarantees, ergonomics, open-source Components, and a growing Enterprise surface.

What This Signals

The Series B signals that investor attention is moving beyond code-generation interfaces toward the systems that make generated software trustworthy. Agents can create an application quickly, but customers still experience every outage, data inconsistency, security mistake, and cost surprise at human speed. The infrastructure beneath the agent remains accountable even when the author is synthetic.

For developers, the practical question is whether a backend platform can remove complexity without hiding the controls needed at scale. For investors, the question is whether the next durable developer platform is a better database, a better cloud abstraction, or a set of agent-friendly guarantees that makes those categories blur together. Convex is using $57M to argue for the third answer.

The company is actively hiring as it pushes that thesis into larger workloads and more demanding organizations. The round buys Convex time and talent, not victory. What happens next will be measured less by how much code agents can produce and more by how confidently teams can let that code run.