SiMa.ai

SiMa.ai, founded in 2018 by semiconductor veteran Krishna Rangasayee (ex-EVP/GM Xilinx, ex-COO Groq), develops what it calls a 'Physical AI' platform: purpose-built silicon and software for running AI inference at the edge without a cloud tether. Its MLSoC (Machine Learning System-on-Chip) entered mass production in 2023; the second-generation Modalix chip targets multimodal generative AI workloads on devices like drones, cameras, and robots. The full stack — branded SiMa.ai ONE — pairs Modalix silicon with the Palette SDK, Palette Edgematic development environment, and LLiMa software runtime. SiMa.ai has raised $355M in total, most recently an $85M round led by Maverick Capital in August 2025. A further Series C (undisclosed amount, reported at a $1.4B target valuation) closed in April 2026.

As AI moves from the cloud to cameras, robots, and vehicles, the bottleneck shifts to power-efficient, programmable edge silicon. SiMa.ai's software-centric MLSoC approach — where the same workload runs unchanged across chip generations — reduces integration friction that has historically held back edge AI deployments in automotive and industrial programs.

sima.ai Edge AIUSAprivate

At a glance

Total raised

$355M

last round series c

Headcount

201-500

approximate · public range

Headquarters

San Jose

USA

Founded

2018

8 years active

Status

Private

Edge AI

Latest valuation

$1.4B

Series C · 2026

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Founding

50% off

/month

91 of 100 spots left

Full

$10

/month

Available once all 100 Founding Member spots are claimed.

Get full access

Already subscribed? Sign in →

Funding history

Total raised

$355M

Rounds tracked

Latest stage

Series C

Latest valuation

$1.4B

Date	Stage	Amount	Valuation	Lead investors
Apr 8, 2026	Series C	—	$1.4B	—
Aug 6, 2025	Series C	$85M	—	Maverick Capital
Apr 4, 2024	Series B (extension)	$70M	—	Maverick Capital
Jun 13, 2023	Series B	$13M	—	VentureTech Alliance
Oct 26, 2022	Series B	$37M	—	MSD Partners
May 1, 2022	Series B	$30M	—	Fidelity Management & Research Company
May 1, 2021	Series B	$80M	—	Fidelity Management & Research Company
May 12, 2020	Series A	$30M	—	Dell Technologies Capital
Aug 1, 2019	Seed	$6M	—	—
Jan 1, 2019	Seed	$4M	—	—

How they make money

Hardware + recurring servicesPattern

SiMa.ai's primary revenue comes from selling its MLSoC silicon — initially the gen-1 chip that entered mass production in 2023, and now the second-generation Modalix family — into automotive, robotics, industrial, and defense OEM programs. Hardware ships as bare silicon and in module/PCIe board form factors, with customers (typically tier-1 automotive suppliers, industrial machine-vision OEMs, drone manufacturers) purchasing in volume once they complete an integration design cycle. Because edge AI programs commit to a silicon platform for 3–5 year product lifecycles, each design win generates a recurring volume tail. The recurring services dimension comes from the software stack (Palette SDK, Palette Edgematic cloud development environment, LLiMa runtime) and customer success / professional-services engagement required to onboard and maintain complex ML pipelines in safety-critical applications. While the SDK is currently offered as part of the platform sale rather than a separate subscription, the company's stated "software-centric" positioning and the Edgematic cloud tool suggest a trajectory toward attaching recurring software revenue — and eventually SDK licensing fees — to each hardware customer as the installed base grows.

Generated by editorial-sourced · June 4, 2026Analytical summary · not investment advice

Founder backgrounds

Krishna Rangasayee

Founder & CEO

Krishna Rangasayee founded SiMa.ai in November 2018 after a 30-year career spanning Cypress Semiconductor, Altera Corporation, Xilinx, and Groq. He spent 18 years at Xilinx, rising to Executive Vice President and GM of the overall business, where he helped grow revenues to $2.5B at ~70% gross margin. Immediately before founding SiMa.ai he served as COO of Groq, giving him direct experience at a prior AI chip startup. He holds a B.E. from the National Institute of Technology Warangal, an MSEE from Mississippi State University, and completed the executive management program at Stanford Graduate School of Business. Rangasayee holds 25+ international patents and has served on the boards of public and private companies.

AI bull / bear

Bull case

SiMa.ai occupies the most defensible position in edge AI silicon: a full-stack "software-centric" platform where the Palette SDK and LLiMa runtime abstract the hardware, so customers port workloads once and re-use them across chip generations. That model mirrors what Xilinx did with FPGAs — exactly the playbook CEO Krishna Rangasayee spent 18 years executing. The Modalix MLSoC is purpose-built for the power-envelope (sub-5W) that cloud-class accelerators cannot meet, targeting the fast-growing robotics, automotive ADAS, and industrial machine-vision markets where inference at the edge is a hard requirement, not a preference. $355M raised at a reported $1.4B valuation with marquee investors (Fidelity, Maverick, Point72, Lip-Bu Tan) and an IDC Innovator designation signals broad institutional validation. The long-term moat is the software ecosystem: if SiMa.ai locks in the dominant SDK for edge AI deployment, switching costs rise dramatically even if competing silicon eventually matches performance-per-watt. The second-generation Modalix's multimodal and generative AI capabilities (LLiMa runtime) position the company to capture the next wave of on-device LMM workloads in automotive infotainment and humanoid robotics — both markets poised for significant volume ramps in 2026–2028.

Bear case

SiMa.ai competes against well-resourced rivals across multiple axes simultaneously: Nvidia's Jetson line owns the robotics developer mindshare, Hailo and Kneron target the same low-power vision segment, and Qualcomm's Snapdragon-based automotive platforms are already designed into major OEM pipelines. The edge AI chip market fragments rather than consolidates, and a startup with sub-$100M revenue is unlikely to sustain the full-stack R&D spend required to keep pace with companies that earn billions per quarter. The multi-tranche Series B structure (five tranches 2021–2024) suggests the company struggled to hit milestones that would unlock a clean new round — a pattern that can reflect slower-than-expected design-win conversion in the long-cycle automotive business. At the silicon layer there is existential pressure: hyperscalers (Nvidia, Google, Amazon) are steadily pushing capable inference chips toward the edge, and if cloud-connected inference economics improve enough, the "no cloud tether" value proposition weakens. SiMa.ai's chip is fabless and manufactured on a leading-edge node, exposing it to the same wafer-allocation risks as all fabless players — a constraint that becomes acute if a ramp demand spike collides with tight foundry capacity.

Generated by editorial-sourced · June 4, 2026Refreshes quarterly · see disclosures

In the news

SiMa.ai · last 30 days

No articles ingested yet for SiMa.ai. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.

Watch

Leadership

Krishna Rangasayee

Founder & CEO

Harry Kroeger

President, Sales and Automotive

—

Gopal Hegde

Head of Engineering and Operations

—

Azfar Hasib

Head of People and Culture

—

Products

Modalix MLSoC
Second-generation machine-learning system-on-chip built for multimodal generative AI at the embedded edge. Targets robotics, cameras, drones, automotive ADAS, and industrial automation, delivering high performance-per-watt inference without cloud connectivity. Ships as standalone silicon and in SoM, PCIe HHHL, and PCIe Dual M.2 board form factors.
SiMa.ai ONE Platform
Integrated Physical AI platform combining Modalix silicon with SiMa.ai's full software stack (Palette SDK, Palette Edgematic, LLiMa). The software-centric design lets customers port workloads once and run them across chip generations, reducing re-integration effort for automotive and industrial customers.
Palette SDK & Edgematic
Software development kit and cloud-accessible development environment (edgematic.sima.ai) for compiling, profiling, and deploying neural networks on SiMa.ai silicon. Supports standard ML frameworks and is the primary tool for onboarding new edge AI applications to the MLSoC family.
LLiMa
Software runtime layer enabling large language and multimodal model inference on SiMa.ai edge silicon. Part of the SiMa.ai ONE stack, it extends the platform's reach from vision-only workloads to generative AI use cases on constrained edge hardware.