MemTensor (记忆张量) Raises Seed Round Led by Oriental Fortune Capital

MemTensor (记忆张量) Closes Seed Round

Shanghai-based AI company MemTensor (记忆张量) has raised a seed round led by Oriental Fortune Capital. The funding was announced on September 4, 2026. Additional investors in the round include Futeng Capital, Xinshang Capital (新尚资本), and Shanghai Yaohong Venture Capital.

The specific funding amount was not disclosed.

What MemTensor Does

MemTensor was founded in November 2024 and is incubated by the Shanghai Institute of Algorithm Innovation. The company's chief scientific advisor is a Chinese Academy of Sciences academician.

MemTensor builds MemOS, an open-source memory operating system for large language models and AI agents. MemOS is designed to let models produce, schedule, update, forget, and audit long-term memory as a managed system resource. This approach contrasts with methods that rely on ever-longer context windows or retrieval augmentation.

By treating memory as an operating-system-level resource, MemTensor aims to give AI systems a structured way to manage what they retain, revise, and discard over time. The company is extending its stack upward into memory-native foundation models, positioning its work around three stated goals: low cost, low hallucination, and high generalization.

The Round

• Round type: Seed
• Amount: Not disclosed
• Currency: CNY
• Announced: September 4, 2026
• Lead investor: Oriental Fortune Capital
• Other investors: Futeng Capital, Xinshang Capital (新尚资本), Shanghai Yaohong Venture Capital

The investor group combines a lead backer in Oriental Fortune Capital with several additional China-based venture firms, giving the round a mix of institutional participation.

Why It Matters

Memory management is an emerging layer in the AI stack. As agents and LLM applications move from single-turn interactions to persistent, multi-session workflows, the question of how models store and retrieve long-term information becomes a systems problem rather than just a prompting problem.

MemTensor's bet is that memory should be handled by an operating system abstraction — with scheduling, updates, forgetting, and auditing as first-class operations — rather than by expanding context windows or bolting on retrieval pipelines. If that thesis holds, MemOS could become infrastructure that other AI applications build on.

The open-source nature of MemOS is also notable. Open-source distribution can drive adoption and standardization, which in turn can create commercial opportunities in support, hosted services, and enterprise deployments.

Company Background

• Founded: November 2024
• Headquarters: Shanghai
• Incubation: Shanghai Institute of Algorithm Innovation
• Scientific advisor: Chinese Academy of Sciences academician
• Core product: MemOS, an open-source memory operating system
• Direction: Memory-native foundation models

Acquisition Angle

For acquirers, MemTensor offers exposure to a specific and increasingly relevant layer of the AI stack:

• Infrastructure positioning: Memory as a managed system resource, not an application feature.
• Open-source footprint: MemOS provides a distribution channel and developer mindshare.
• Research backing: Incubation by the Shanghai Institute of Algorithm Innovation and an academician scientific advisor.
• Investor validation: A seed round led by Oriental Fortune Capital with three additional venture participants.
• Roadmap extension: Movement from memory OS toward memory-native foundation models.

Strategic buyers in AI infrastructure, model development, or enterprise AI platforms could view memory management as a complementary capability to existing model and agent offerings. The company's early stage means any acquisition interest would likely center on team, research direction, and the MemOS codebase rather than current revenue.

What to Watch

With the seed round closed, the key signals to monitor are adoption of MemOS in the open-source community, progress toward memory-native foundation models, and whether the company converts its research positioning into commercial deployments. The involvement of a Chinese Academy of Sciences academician as chief scientific advisor suggests a research-heavy orientation that may shape both hiring and product priorities.