Rivos made headlines in 2022 after Apple filed a trade secrets lawsuit against him, accusing Rivos of hiring dozens of Apple engineers and using confidential information to develop chips rivaling those from the iPhone manufacturer.
Rivos denied the allegations and sued Apple for unfair competition. Apple ultimately settled its lawsuit in February. Around the same time, it ended a separate dispute with several of the Apple engineers Rivos had hired.
Now, with the legal drama behind it, Rivos is doubling down on its efforts to commercialize its chipset technology, CEO Puneet Kumar told TechCrunch.
“Rivos was founded with a mission to build cutting-edge, energy-efficient, high-performance chips,” Kumar said. “We are excited to target customers who are creating data-driven solutions. »
A substantial new tranche of funding will help fund these efforts.
Rivos announced Tuesday that it has raised more than $250 million in an expanded and oversubscribed Series A round led by Matrix Capital Management with participation from chip giants including Intel (via its corporate VC division) and MediaTek. Other backers included Cambium Capital, Hotung Venture Group, Walden Catalyst, Dell Technologies Capital and Koch Disruptive Technologies.
It’s quite a turnaround for Rivos, which was founded in 2021 and about a year ago was struggling to raise money from investors and hire employees in the shadow of the Apple lawsuit. In August, Rivos laid off nearly two dozen employees, or 6% of its workforce at the time, and was forced to delay a planned $400 billion Series A fundraising, The Information reported at the time.
A custom server chip
Rivos’ long-term goal, Kumar said, is to build chips primarily for servers that can handle intensive data analytics and AI workloads, including generative AI workloads. .
“We are targeting customers who are building data-driven solutions, for example those who use generative AI and data analytics to make decisions,” Kumar said. “Many companies target such markets; Rivos supports the intense hardware demands of the AI models and analytics that will remake the business.
Rivos’ first chipset is built on RISC-V, the open standard instruction set architecture (ISA).
ISAs are a technical specification at the base of each chip, describing how the software controls the chip’s hardware. For general purpose computing, chip design teams typically obtain an existing ISA license from an incumbent (e.g. Arm or Intel). But RISC-V presents an open, royalty-free alternative.
Rivos’ chip includes what Kumar describes as a “parallel data accelerator” to speed up calculations related to AI and big data, essentially a GPU designed for purposes beyond graphics processing. It was made using TSMC’s 3nm manufacturing process. In chip manufacturing, “process” refers to the size of the smallest component that can be integrated onto a chip.
These 3 nm are considered close to the tip. While Qualcomm, MediaTek, Nvidia and AMD, among others, are expected to use TSMC’s process for their upcoming chip families, Apple was the only company to use it in 2024 in its M3 series of chipsets.
In addition to building the chip, Rivos is working on standalone data center hardware based on the Open Compute Project modular standard, which will effectively serve as a plug-and-play chip package. And that creates a “firmware-to-app” software stack for programming the chip, Kumar said.
“Customer workloads can be easily deployed on our more efficient hardware, while using their existing models and databases, giving them an immediate advantage,” added Kumar.
Rivos, which is currently in pre-revenue, plans to make money by charging its customers – primarily large data center operators – for its hardware and complementary software solutions. David Goel, an early investor, said Rivos’ “low-friction” adoption pipeline is a key differentiator in the cutthroat chip market.
“The Rivos team has skillfully integrated the revolutionary new RISC-V architecture with an inventive accelerator, bringing this vision to life,” Goel told TechCrunch. “Their prototype chip provides a compelling demonstration of their unique capability. »
But is it different enough?
Strong competition
One of Rivos’ potential customer segments, large technology companies, are rushing to develop their own in-house chips for AI and big data analytics as the generative AI boom continues.
Google is working on its fifth-generation TPU and recently unveiled Axion, its first chip dedicated to running models. Amazon has several custom chip families under its belt. Last year, Microsoft entered the fray with the Azure Maia AI accelerator and the Azure Cobalt 100 processor. And Meta is moving forward with its own creations.
Meanwhile, dozens of startups are aiming for a share of the custom data center chip market that could reach $10 billion this year and double by 2025.
Groq, a company developing chips to run AI models faster than conventional hardware, recently formed a new business unit focused on enterprise applications and use cases. AI hardware startup Tenstorrent, led by engineering luminary Jim Keller, is looking to integrate its chipsets into data centers. And Rebellions, a South Korean fabless AI chip company, has raised hundreds of millions of dollars in capital to accelerate production of its Atom chip for data centers.
But Nvidia, the dominant force in chips today, is proving difficult to topple.
Nvidia briefly became a $2 trillion company this year, riding high on demand for its GPUs for AI training. Wells Fargo Equity Research estimates that Nvidia has a 98% market share in data center GPUs, and that the company’s data center business grew more than 400% in the fourth quarter of 2023, as Nvidia is building a new unit to design custom chips for cloud computing companies and others.
Given the fierceness of the competition – and the chilling effect that Nvidia’s supremacy has had on funding potential rivals – this has been difficult for some custom server chip newcomers.
Graphcore, whose valuation was reportedly slashed by $1 billion after a failed deal with Microsoft, said a few months ago that it planned job cuts due to the “extremely challenging” macroeconomic environment. difficult “. Habana Labs, the AI chip company owned by Intel, laid off about 10% of its workforce last year. Also last year, SiFive – like Rivos, a RISC-V startup – laid off 20% of its workforce and discontinued its core product line.
So will Rivos fare better? Maybe.
Kumar wouldn’t talk about customers, and Rivos’ chip isn’t expected to reach mass production until next year. But with 375 employees and hundreds of millions of dollars in the bank, Kumar said Rivos is well-positioned to expand manufacturing and double down on platform and software engineering efforts.
“Rapid changes in generative AI and the merger with the data analytics stack make it essential that accelerators are easy to program and debug, and that data can flow seamlessly between the processor and the accelerator” , Kumar said. “Rivos addresses this need with our “recompile, not rethink” approach. »
techcrunch
