SiMa.ai, a San Jose-based chipmaker that builds silicon for robots, drones, and cameras rather than cloud data centers, closed a $150 million Series C round on September 28, 2026, pushing its valuation to $1.45 billion, according to TechCrunch and the company’s own press release. The round, co-led by Fidelity Management & Research Company and Amplify, brings SiMa.ai’s total funding to $500 million and marks a roughly 51% jump from the $960 million valuation it held after its Series B just over a year earlier.
The timing matters. While most of the AI funding headlines in 2026 have gone to frontier model labs and hyperscale data center buildouts, SiMa.ai’s raise is a bet on the opposite end of the spectrum: AI that runs directly on physical devices, with no round trip to the cloud. The company calls this “physical AI,” and its new backers are wagering that the next big compute market is not another GPU cluster but the chips inside humanoid robots, delivery drones, and self-driving cars.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
SiMa.ai’s $150 Million Round, by the Numbers
SiMa.ai confirmed the financing in a press release published on its own site, with TechCrunch, SiliconANGLE, The Next Web, and Unite.AI all independently reporting the same core figures on September 28. The $150 million Series C values the eight-year-old company at $1.45 billion, officially vaulting it into unicorn territory. SiMa.ai says its cumulative funding since founding now stands at $500 million, meaning this single round accounts for roughly 30% of everything the company has ever raised.
That growth curve is steep by any measure. SiMa.ai’s Series B, reported by TechCrunch citing PitchBook data, closed in July 2025 at $85 million and valued the company at approximately $960 million. In just over 14 months, SiMa.ai nearly doubled its valuation and raised almost twice as much capital in a single round as it had in the prior one. For a semiconductor startup, that kind of acceleration usually signals either a breakout product milestone or a land grab among investors trying to get into a hot category before prices rise further. SiMa.ai’s messaging points to the latter: an investor rush into physical AI silicon before the market matures.
| Metric | Series B (July 2025) | Series C (Sept. 28, 2026) |
|---|---|---|
| Round size | $85 million | $150 million |
| Post-round valuation | ~$960 million | $1.45 billion |
| Cumulative funding | $355 million | $500 million |
| Lead investors | Not fully disclosed in available reporting | Fidelity Management & Research Company, Amplify |
| Stated use of funds | Product scaling | Palette Neat software, next-gen Physical AI silicon |
Who Is Backing SiMa.ai’s Edge AI Bet
Fidelity Management & Research Company and Amplify served as co-lead investors on the Series C, according to SiMa.ai’s press release and matching TechCrunch and SiliconANGLE coverage. The round also drew participation from Alter Venture Partners, Dell Technologies Capital, Maverick Capital, ND Capital, Point72, and StepStone Group, with some reports additionally naming AllianceBernstein, Baron Capital, J.P. Morgan, and the State of Michigan among participants. Coverage varies slightly on the complete investor list, but Fidelity and Amplify as co-leads is consistent across sources.
That mix is notable. Dell Technologies Capital’s involvement signals a strategic enterprise-hardware interest, while the presence of asset managers like Fidelity, AllianceBernstein, and Baron Capital suggests institutional investors are now treating physical AI chip startups as a distinct, fundable category rather than a side bet within broader AI portfolios. A state-level investor, Michigan, also appearing on the list points to regional economic-development interest in onshoring advanced chip and robotics supply chains, a theme that has run through U.S. industrial policy for several years.
From Groq’s COO to Founder: Who Runs SiMa.ai
SiMa.ai was founded in 2018 by Krishna Rangasayee, who serves as the company’s CEO. Before starting SiMa.ai, Rangasayee was chief operating officer at Groq, the AI-chip company now known for its language-model inference hardware, according to TechCrunch’s reporting on the round. That background is relevant context: Rangasayee spent his pre-SiMa.ai career inside a company built around specialized AI silicon, and he has spent the years since steering SiMa.ai toward a different slice of the same broader trend, moving inference out of the data center and onto the device itself.
Available reporting does not identify additional named co-founders or provide a more detailed leadership biography beyond the Groq connection, so this article does not speculate further on SiMa.ai’s executive bench beyond what sources confirm.
What “Physical AI” Actually Means
SiMa.ai and its investors use the term “physical AI” to describe machine-learning workloads that run on embedded devices operating in the real world, as opposed to the generative AI chatbots and agents that dominate headlines. The company’s press materials and coverage from TechCrunch frame the category as encompassing humanoid robots, medium- to high-end drones, automotive advanced driver-assistance systems (ADAS), and AI-enabled vehicle cockpits. The common thread is latency and power: a drone avoiding an obstacle or a robot catching a dropped object cannot afford the round-trip delay of sending sensor data to a cloud server and waiting for a response, and it cannot carry a power-hungry data-center GPU on board either.
That constraint is exactly what SiMa.ai says its architecture is built to solve: energy-efficient, purpose-built silicon paired with a software stack, designed to run inference locally rather than transmitting data back to the cloud repeatedly, as TechCrunch’s reporting on the company describes it. It is a narrower, more specialized market than general-purpose AI acceleration, but one that scales with every new category of autonomous hardware that reaches commercial deployment.
Inside the Stack: MLSoC, Modalix, and Palette Neat
SiMa.ai’s current shipping hardware is built around its Modalix MLSoC, a machine-learning system-on-chip described in available coverage as energy-efficient embedded compute for on-device inference, offered in at least two deployable forms: a PCIe Card for host-based systems and a System-on-Module (SoM) for integrated edge appliances. Modalix production shipments began in August 2025, and the Modalix SoM later won the Edge AI + Vision Alliance 2026 Product of the Year award. Reporting on the Series C does not provide a complete independently verified specification sheet for the current MLSoC, such as process node, memory bandwidth, or an exact TOPS (trillions of operations per second) rating, so this article does not state those figures.
On the software side, SiMa.ai is directing part of the new capital toward Palette Neat, SiMa.ai’s agentic software environment for physical AI development. The company’s press release names scaling this software as one of the two primary uses of the Series C proceeds, alongside hardware. The current hardware is called Modalix MLSoC, described in company materials as a machine learning system on chip aimed specifically at embedded robotics and other physical AI applications, and it is already shipping and deployed in systems such as the Wobot.ai partnership. SiMa.ai is also developing next generation hardware, including machine learning IP, chiplets, and SoCs targeting 1,000 dense TOPS, planned for the first half of 2028.
The 1,000-TOPS Roadmap
SiMa.ai said the new financing will support development of next-generation physical AI silicon capable of up to 1,000 dense TOPS of compute, targeted for the first half of 2028, according to the company’s own announcement. That figure is a development target for future silicon, not a specification of a chip currently shipping to customers, and while SiMa.ai’s own blog puts the power target at under 80 W, the available reporting does not specify the precision format or benchmark methodology behind the number. Coverage also notes SiMa.ai’s broader roadmap includes new machine-learning intellectual property and chiplet designs, suggesting the company is moving toward a modular architecture rather than a single monolithic chip line.
For context on how aggressively AI chip roadmaps are scaling across the industry right now, SiMa.ai’s 1,000-TOPS target arrives in the same window as AMD’s 256-core EPYC 9006 “Venice” data-center chip and continued expansion of Nvidia’s silicon roadmap, underscoring that compute targets are rising across every tier of the AI hardware stack, from the data center down to the edge device.
SiMa.ai’s Customer Base: 150-Plus Companies and Counting
SiMa.ai says its technology is in use by more than 150 customers across automotive, drone, and robotics markets, a figure cited in company materials and corroborated by industry coverage. Named partners and customers across that reporting include Bosch, Micron, Synopsys, Emerson, TRUMPF, Kontron, AVerMedia, ARK Electronics, Intrinsic, L&T Technology Services, STIGA, and Virya Autonomous Technologies. The application areas described in coverage span humanoid and other robotics (with Modalix specifically positioned for embedded robotics), medium- to high-end drones, automotive ADAS and AI-powered cockpits, and local AI processing in cameras and other edge devices.
That customer roster reads like a cross-section of industrial and automotive supply chains rather than consumer tech, which fits the physical AI thesis: these are companies that build the hardware going into factories, vehicles, and warehouses, not smartphone apps. It also explains why strategic investors like Dell Technologies Capital and customers like Micron have an interest in SiMa.ai’s trajectory beyond pure financial return. Micron’s own memory supply constraints, which this site covered in its reporting on Micron’s 2027 memory supply being largely sold out, mean its chip-customer relationships across the AI hardware ecosystem, including with companies like SiMa.ai, are increasingly strategic rather than purely transactional. Micron Technology also made a strategic investment in SiMa.ai in April 2026 aimed at compute-memory architectures.
Competitive Landscape: SiMa.ai vs. the Edge AI Field
SiMa.ai competes in a crowded field that spans both semiconductor giants and specialized startups. Qualcomm and Nvidia (with its Jetson line) bring massive scale and existing developer ecosystems to edge AI compute. Smaller, more specialized rivals named in coverage of the edge AI space include Hailo, Ambarella, and Mythic, each targeting overlapping but distinct niches within on-device inference. Rangasayee’s former employer, Groq, occupies an adjacent but different lane, focused primarily on cloud inference acceleration for large language models rather than embedded physical AI.
Available reporting on the Series C does not include directly comparable market-share figures or head-to-head benchmark data between SiMa.ai and these competitors, so this article avoids asserting a specific ranking. What differentiates SiMa.ai in its own positioning, according to company statements, is the combination of a purpose-built chip architecture with an integrated software stack (Palette Neat) aimed squarely at physical AI developers, rather than retrofitting general-purpose edge chips for robotics and automotive use cases. For a broader look at how the GPU and AI-chip market outside Nvidia and AMD has evolved, see this site’s earlier coverage of who makes GPUs besides Nvidia and AMD.
| Company | Primary Focus | Target Market |
|---|---|---|
| SiMa.ai | Physical AI SoCs + software (MLSoC, Modalix, Palette Neat) | Robotics, drones, automotive ADAS, cameras |
| Nvidia Jetson | General-purpose edge AI modules | Robotics, embedded AI, industrial |
| Qualcomm | Mobile and automotive AI silicon | Smartphones, automotive cockpits, IoT |
| Hailo | Dedicated edge AI inference chips | Cameras, smart devices, automotive |
| Ambarella | AI vision processors | Cameras, security, automotive vision |
| Groq | Cloud LLM inference acceleration | Data center, not embedded edge |
Why Edge AI Funding Is Accelerating in 2026
SiMa.ai’s raise is not happening in isolation. Industry research tracked by IDC has repeatedly flagged the scale of the shift toward AI computing outside traditional data centers. According to IDC data cited by Computerworld, shipments of AI-enabled PCs were expected to grow from nearly 50 million units in 2024 to more than 167 million units by 2027. The same report, citing IDC and Gartner, found that AI PCs and generative-AI-capable smartphones combined were projected to reach 295 million units by the end of 2024, up from just 29 million units in 2023. Those figures describe consumer AI hardware rather than industrial physical AI specifically, but they point to the same underlying trend driving SiMa.ai’s valuation: AI compute is migrating toward the device, and investors are following it there.
That migration has already reshaped funding patterns elsewhere in the chip sector this year. CScale’s $145 million raise for AI optical networking landed just weeks before SiMa.ai’s round, another example of investors funding the physical infrastructure layer beneath AI rather than the models themselves. Chip M&A activity tells a similar story: AMD’s $8.2 billion acquisition of World Labs, which brought AI researcher Fei-Fei Li on board as chief scientist, was explicitly framed around spatial intelligence and physical-world AI applications, the same thesis underpinning SiMa.ai’s pitch to investors.
Market Impact: What a $1.45 Billion Valuation Signals
A 51% valuation jump in roughly 14 months, from $960 million to $1.45 billion, places SiMa.ai among a small group of AI-chip startups commanding unicorn-plus valuations without having gone public. That matters for the broader semiconductor funding environment in two ways. First, it signals late-stage institutional capital, the kind represented by Fidelity, AllianceBernstein, Baron Capital, and J.P. Morgan in this round, is now comfortable underwriting pre-IPO physical AI chip bets at billion-dollar-plus valuations, not just early-stage venture money. Second, it raises the bar for competing edge AI startups seeking their own funding rounds, since later entrants will likely be priced against SiMa.ai’s $1.45 billion mark as a reference point.
It also lands against a backdrop of broader AI-chip revenue growth across the industry. Broadcom’s AI chip revenue hit $16.7 billion in a recent quarter, up 221% year over year, evidence that AI silicon demand is climbing across both data-center and edge categories simultaneously. SiMa.ai’s much smaller but fast-growing valuation sits at the edge-device end of that same demand curve. Similarly, CoreWeave’s rollout of Nvidia’s Vera AI agent CPU shows compute providers racing to serve autonomous workloads at every layer of the stack, not just inside hyperscale clusters.
Historical Context: Eight Years From Founding to Unicorn
SiMa.ai’s path from its 2018 founding to a $1.45 billion valuation in 2026 tracks a broader arc in AI chip history. The company launched during a period when “edge AI” was still a niche engineering concern, years before generative AI pushed AI hardware into the mainstream investment conversation. Rangasayee’s bet, drawn from his time at Groq, was that specialized silicon would outperform general-purpose chips for specific AI workloads, a thesis that proved out first in cloud inference (via companies like Groq itself) and is now proving out at the edge via SiMa.ai.
That eight-year runway also reflects how capital-intensive and slow-moving semiconductor development remains, even amid an AI funding boom that has compressed timelines elsewhere in software. A chip company needs multiple silicon tape-outs, extensive customer qualification cycles in regulated industries like automotive, and years of software stack maturation before a funding round like this one becomes possible. SiMa.ai’s Series C is as much a validation of that multi-year grind as it is a bet on where physical AI goes next. The same pattern shows up in Arm’s push into AGI-focused CPU designs aimed at AMD and Nvidia’s data center dominance, where years of architecture work are now colliding with a sudden surge of investor attention.
Predictions: Where Physical AI Chip Funding Goes From Here
Based on the trajectory of this round and the broader funding environment, several outcomes look likely over the next 12 to 18 months.
- More institutional asset managers, following Fidelity, AllianceBernstein, and Baron Capital’s lead, will likely enter pre-IPO rounds for edge and physical AI chip startups, treating the category as investable in its own right rather than a venture-only niche.
- Competitive pressure will push rivals named in this space, including Hailo and Ambarella, toward their own larger funding rounds or acquisition talks, as strategic buyers look to avoid falling behind SiMa.ai’s valuation benchmark.
- SiMa.ai’s Modalix MLSoC is already shipping and deployed in real world robotics, automotive, and drone systems, so it is already under public scrutiny. The company’s next generation silicon, targeting 1,000 dense TOPS through machine learning IP, chiplets, and SoCs, is planned for the first half of 2028, and will face its own scrutiny once those chiplets and SoCs ship.
- Automotive ADAS and humanoid robotics will remain the two fastest-growing demand categories for physical AI silicon, given the customer names already disclosed across SiMa.ai’s 150-plus customer base.
- Expect continued consolidation pressure in the edge AI chip space, echoing the pattern already seen in AMD’s acquisition of World Labs, as larger semiconductor players look to buy rather than build physical AI capabilities.
What It Means for Developers and Enterprise Buyers
For engineering teams building robotics, drone, or automotive products, SiMa.ai’s raise is a signal that the physical AI chip category now has staying power and capital behind it, reducing the platform risk of committing to its MLSoC and Modalix MLSoC hardware for multi-year product cycles. The Palette Neat software investment also matters directly to developers, since SiMa.ai is explicitly positioning it as an agentic development environment rather than a bare-metal SDK, suggesting the company wants to lower the barrier to building physical AI applications rather than requiring deep embedded-systems expertise from every customer team.
Enterprise buyers evaluating edge AI vendors should treat this round as one data point among several, alongside Nvidia’s Jetson ecosystem maturity, Qualcomm’s automotive design wins, and specialized players like Hailo and Ambarella, rather than a signal that SiMa.ai has definitively won the category. Procurement teams in regulated industries like automotive typically run multi-year qualification processes regardless of a vendor’s funding news, so this raise extends SiMa.ai’s runway to compete for those design wins rather than guaranteeing it secures them.
Frequently Asked Questions
What is SiMa.ai?
SiMa.ai is a San Jose-based semiconductor company, founded in 2018 by CEO Krishna Rangasayee, that builds chips and software for running AI directly on physical devices such as robots, drones, and cameras, rather than in cloud data centers.
How much did SiMa.ai raise in its Series C round?
SiMa.ai raised $150 million in its Series C round, announced September 28, 2026, valuing the company at $1.45 billion and bringing its total funding to $500 million, according to the company’s press release and TechCrunch’s reporting.
Who led SiMa.ai’s Series C funding round?
Fidelity Management & Research Company and Amplify co-led the round, with participation from investors including Alter Venture Partners, Dell Technologies Capital, Maverick Capital, Point72, and StepStone Group, among others named across coverage.
What does “physical AI” mean?
Physical AI refers to machine-learning systems that run directly on embedded devices operating in the real world, such as robots, drones, and self-driving vehicle systems, as opposed to cloud-based generative AI models and chatbots.
What products does SiMa.ai make?
SiMa.ai’s current hardware platform is its MLSoC, a machine-learning system-on-chip for embedded inference. Its current hardware also includes Modalix MLSoC, a machine learning silicon family for embedded robotics that is already shipping and deployed in robotics, automotive, and drone systems. The company’s software platform, Palette Neat, supports physical AI development.
Who are SiMa.ai’s customers?
SiMa.ai says it has more than 150 customers, with named partners including Bosch, Micron, Synopsys, Emerson, and TRUMPF, spanning automotive, drone, and robotics applications.
Who are SiMa.ai’s main competitors?
SiMa.ai competes with large semiconductor companies including Qualcomm and Nvidia (via its Jetson platform), as well as specialized edge AI chip makers such as Hailo, Ambarella, and Mythic.
What was SiMa.ai’s valuation before this round?
SiMa.ai was valued at approximately $960 million following an $85 million Series B round in July 2025, according to TechCrunch’s reporting citing PitchBook data.
