EnCharge AI Analog In-Memory Computing Accelerators

EnCharge AI offers analog in-memory computing hardware and software for AI from edge to cloud. The company launched in 2022 from research at Princeton, and CEO and co-founder Naveen Verma leads it. Offices are in Santa Clara, California and Bengaluru, India.

Its charge-domain computation uses intrinsically precise metal capacitors to avoid the signal-to-noise limits of analog processing, which the company says is proven through 5 generations of designs. Products span chiplets, ASICs and standard PCIe cards, aimed at running advanced AI models within tight power, space and cost limits.

• Analog in-memory compute: charge-domain computation that reduces data movement.
• Precise capacitors: intrinsically precise metal capacitors address analog signal-to-noise limits.
• Product formats: chiplets, ASICs and standard PCIe cards.
• Efficiency: claimed 20x TOPS per watt and 9x TOPS per mm2.
• Programmability: support for varied AI model types and operators within user workflows.
• Silicon maturity: 350M+ chips shipped, 150+ patents granted and 300+ technical publications.

An electronics maker can fit EnCharge PCIe cards into workstations or edge servers for on-premises AI. An industrial equipment vendor can embed chiplets or ASICs for local inference with low power. A device brand can evaluate on-device AI features without cloud dependence.

Buyers can run AI models on devices and servers with far less energy than conventional digital accelerators. The company claims 20x higher efficiency in TOPS per watt, 9x higher compute density and 10x lower total cost of ownership, based on silicon-measured MACs. Chiplet, ASIC and PCIe card formats give integration options for device makers.

AI workloads are moving onto laptops, workstations, industrial equipment and edge servers where power and cost budgets are tight, so energy per inference is a key purchase criterion. Vietnamese and Southeast Asian PC, embedded and industrial equipment makers may look for efficient accelerators to add on-device AI.

Launched in 2022; Series B of more than USD 100 million announced 13 February 2025, led by Tiger Global, bringing total funding to more than USD 144 million. Investors include Samsung Ventures, In-Q-Tel, RTX Ventures and HH-CTBC. Offices in Santa Clara, California and Bengaluru, India.