Tech / AI

Graphon Labs Raises $35 Million for Persistent AI Memory

The company reports 81.7 percent on a Cambridge multimodal memory benchmark and is selling the memory layer as enterprise, developer and on-device products.

Nico ReedAI editorial voice · 1 min read

A data center corridor. Graphon's memory layer is built to be queried across an enterprise's information wherever it sits. Photo: GAP / public domain, via Wikimedia Commons.

Summary

Graphon Labs raised a $35 million Series A to scale persistent memory infrastructure for AI agents.

Tap a passage. React to the part that matters.

Graphon Labs raised $35 million to build memory that AI agents keep between sessions, the company announced.

Graphon Labs announced a $35 million Series A on November 10, 2026, led by RRE Ventures, bringing total funding to $43.3 million for its persistent memory infrastructure for AI agents across enterprise, developer and on-device products. [Unite.AI — Graphon Labs raises $35M Series A to scale persistent AI memory (November 10, 2026)]

Graphon calls the approach Persistent Intelligence: an agent that can continuously access, understand, remember and build on earlier information instead of starting each session cold. The memory spans video, audio, images and text, and the company reports 81.7 percent accuracy on ATM-Bench, a University of Cambridge benchmark for question answering across those formats. The architecture is sold three ways: Graphon Enterprise, governed memory that can be hosted by Graphon, run in a Graphon- or customer-managed cloud, or installed on-premises; Graphon Developer, for builders working across data types without rebuilding the context stack each time; and Graphon On-Device Memory, aimed at phones, laptops, cameras and wearables. Chief executive Arbaaz Khan says the capital moves the company from early experimentation to scaled enterprise deployments. [Unite.AI — Graphon Labs raises $35M Series A to scale persistent AI memory (November 10, 2026)]

The company is young even by AI-startup standards. Graphon was founded in July 2025 in San Francisco by Khan, chief technology officer Clark Zhang and chief operating officer Deepak Mishra, and emerged from stealth on May 14, 2026, with an $8.3 million seed round led by Novera Ventures, naming South Korean conglomerate GS as an early enterprise customer. The Series A adds HP, Energy Impact Partners, Hitachi Ventures and Gaia VC alongside RRE. The commercial question is the same one facing every memory-layer startup: whether persistent memory becomes infrastructure enterprises buy separately, or a feature the model platforms absorb. Graphon's $43.3 million buys it time to prove the first answer. [Unite.AI — Graphon Labs raises $35M Series A to scale persistent AI memory (November 10, 2026)]

The conversation