MemTensor Raises Funding for AI Memory
MemTensor completed a Pre-A+ financing round to develop AI memory infrastructure and memory-native models.
Long-running AI agents need memory, not just bigger context windows.
What happened
Chinese AI startup MemTensor completed a Pre-A+ financing round led by Oriental Fortune Capital, with participation from state-backed and local investors. The company is building AI memory infrastructure and memory-native models, including work around systems for storing, retrieving and reasoning over context.
The round size was not clearly disclosed, so the financing should be treated as an early infrastructure signal rather than a headline mega-round.
Why it matters
As AI systems move from one-off chat into agents, they need to remember tasks, preferences, past decisions and external context across sessions. That creates a new technical layer around memory management, retrieval and reliability.
Without better memory, agents become brittle: they repeat work, lose context and struggle with long workflows.
The bigger picture
AI infrastructure is fragmenting into specialised layers: compute, data, evaluation, security and memory. MemTensor sits in one of the less glamorous but increasingly important parts of that stack.