Engram Raises $98 Million to Cut AI Token Costs

On June 23, Engram, an eight-month-old AI-memory startup, announced $98 million financing. The round was led by General Catalyst, Kleiner Perkins, Sequoia, and OpenAI co-founder Andrej Karpathy, now at Anthropic. Engram markets its technology as a “learned memory” layer that recalls organization-specific workflows, enabling AI models to answer queries with far fewer tokens.

CNBC

Developers have increasingly deployed large language models whose token consumption drives up operational expenses. Recent industry observations note that newer, more sophisticated models are “pricier than previous iterations,” challenging the expectation that scale automatically reduces cost. Companies therefore seek mechanisms to preserve model performance while curbing token usage.

CNBC

Engram’s architecture stores organization-specific knowledge in a persistent memory module. When a user query arrives, the system retrieves relevant context from this memory, allowing the underlying language model to generate answers with reduced token generation. Engram claims its models can match or outperform frontier labs while using up to 100 times fewer tokens, delivering “orders of magnitude cheaper output.”

CNBC

• Investors: General Catalyst, Kleiner Perkins, Sequoia, Andrej Karpathy (OpenAI co-founder, now at Anthropic).

CNBC

• Customers: Microsoft, Notion, and legal-AI startup Harvey have signed on to pilot Engram’s technology.

Engram raises $98 million to cut token costs

• Funding amount: $98 million.

Engram raises $98 million to cut token costs

• Company age: 8 months (founded October 2025).

Nvidia, Micron and AMD weakness raises AI memory-chip worry

• Employee headcount: ?13.

Nvidia, Micron and AMD weakness raises AI memory-chip worry

• Token-efficiency claim: up to 100× fewer tokens than comparable models.

CNBC

• Reported customers: Microsoft, Notion, Harvey.

Engram raises $98 million to cut token costs

Engram’s spokesperson said the $98 million will be used to expand compute resources and hire talent to scale the memory-augmented models. Kleiner Perkins partner Leigh Marie Braswell observed that data volumes and associated costs are surging, and that Engram’s approach could map an organization’s knowledge to deliver substantially cheaper AI output. The company’s messaging reflects the market pressure to curb rising AI operating expenses.

CNBC

• “You've got this explosion of data, explosion of cost,” — Leigh Marie Braswell, Partner, Kleiner Perkins

CNBC

• “Engram comes in and basically maps out your organization and offers orders of magnitude cheaper output.” — Leigh Marie Braswell, Partner, Kleiner Perkins

CNBC

If Engram’s token-reduction claims hold at scale, enterprises could lower AI operating budgets dramatically, making large-model deployments financially viable for a broader set of use cases. The funding round signals investor confidence that memory-centric AI architectures may become a cost-controlling standard in the industry.

CNBC

Engram plans to allocate the $98 million toward expanding its compute capacity, hiring additional engineers, and deepening integrations with its early customers. The company will continue developing its memory-augmented models and scaling its service for enterprise use.

CNBC