SiMa.ai plans next-generation chip shipments for first half of 2028

The $150 million Series C brings reported total funding to $500 million. SiMa.ai plans to use the money to develop new chips and expand its software environment.

The $150 million Series C brings reported total funding to $500 million. SiMa.ai plans to use the money to develop new chips and expand its software environment.
SiMa.ai plans to ship its next-generation hardware in the first half of 2028 for drones, humanoid robots and vehicle systems, Semiconductor Engineering reported on 12 October. The company claims a target of 1,000 dense TOPS for that hardware, distinct from the 50 TOPS it reports for its current chip.
The shipping plan follows the $150 million financing that Semiconductor Engineering reported on 2 October for hardware and software that turn sensor data into real-time action. The 12 October report identifies the financing as a Series C round co-led by Fidelity Management & Research Company and Amplify Partners, and puts SiMa.ai’s total funding at $500 million.
TOPS means trillion operations per second, a measure of computing throughput. “Dense” refers to computation without skipping operations involving zero values. Both performance figures below are company claims, not independent benchmark results; the next-generation figure remains a development target.
| Hardware | Throughput | Status |
|---|---|---|
| Current machine-learning chip | 50 TOPS at under 10 watts | Company-reported performance across vision, language and multimodal-model workloads |
| Next-generation hardware | 1,000 dense TOPS | Company target, with shipments planned for the first half of 2028 |
The report describes SiMa.ai’s current system-on-chip—a chip combining several computing functions—as handling the full processing chain without external accelerator chips. That chain includes capturing and preparing sensor data, converting visual data into a form models can use, and running large multimodal models, which work with more than one type of data.
Semiconductor Engineering describes this work as moving more intelligence into edge devices, which process data close to where it is collected. Turning sensor readings into immediate action increases demands on computing capacity, memory, connectivity, power efficiency and security.
The 12 October report also describes SiMa.ai’s open-source software environment, built around superintelligence (SI) agents—software that can plan and carry out tasks. It compiles models in the ONNX format for the chip and coordinates inference, or running trained models, between the host computer and the chip.
The financing is intended to expand that software environment and develop the next-generation chips. SiMa.ai’s planned shipments in the first half of 2028 target drones, humanoid robotics, advanced driver-assistance systems and vehicle cockpits.

The Justice Department alleges Anibal Alexander Canelon Aguirre helped lead an ATM-theft conspiracy. He was among the Tren de Aragua members sanctioned last week.
3 min

Semiconductor Engineering reported growth of more than 262% from a year earlier, attributing the increase to stronger memory demand, more shipments and higher memory prices.

The planned spending in South Korea and Vietnam targets the circuit boards that connect chips to mainboards, not production of the chips themselves.
1 min