An increasing number of startups are focusing on RISC-V.
In 2022, Rivos made headlines after Apple filed a business lawsuit against it, accusing Rivos of hiring dozens of Apple engineers and using confidential information to develop chips.
The company denied these allegations and counter-sued Apple for unfair competition. Apple eventually settled the lawsuit in February of this year. Around the same time, it also ended separate lawsuits with several Apple engineers hired by Rivos.
Now, with the court drama behind them, Rivos is redoubling its efforts to bring its chipset technology to market, CEO Puneet Kumar told TechCrunch.
"Rivos was founded with the mission to build industry-leading, energy-efficient, high-performance chips," said Kumar. "We are excited to target customers who are building data-driven solutions."
A substantial new round of funding will help support these efforts.
On Tuesday, Rivos announced that it has raised over $250 million in an oversubscribed extended Series A funding round, led by Matrix Capital Management, with participation from chip giants including Intel (through its corporate venture capital division) and MediaTek. Other investors include Cambium Capital, Hotung Venture Group, Walden Catalyst, Dell Technologies Capital, and Koch Disruptive Technologies.For Rivos, which was established in 2021, this marks a significant shift. About a year ago, under the shadow of an Apple lawsuit, Rivos was struggling to raise funds from investors and recruit employees. According to The Information, in August of this year, Rivos laid off nearly 20 employees, representing 6% of its workforce at the time, and was forced to postpone its planned $40 billion Series A funding round.
Advertisement
Custom Server Chips
Kumar stated that Rivos's long-term goal is to build chips for servers capable of handling intensive data analysis and AI workloads, including generative AI workloads.
Kumar said, "Our aim is to attract customers developing data-driven solutions (for example, those leveraging generative AI and data analysis to drive decision-making)."
Rivos's first chip set is built on the open standard instruction set architecture (ISA) RISC-V.
ISA is the foundational technical specification for every chip, describing how software controls the hardware of the chip. For general computing, chip design teams typically license existing ISAs from established companies (such as Arm or Intel). However, RISC-V offers an open, royalty-free alternative.
Rivos's chips feature what Kumar describes as a "data parallel accelerator" to speed up computations related to AI and big data, essentially a GPU designed for purposes beyond graphics processing. It is manufactured using TSMC's 3nm process technology.
Companies such as Qualcomm, MediaTek, Nvidia, and AMD are expected to adopt TSMC's 3nm process in their upcoming chip series, but Apple is the only company to use this process in its M3 chip series by 2024.
In addition to building chips, Rivos is also working on independent data center hardware based on the Open Compute Project's modular standards, which will effectively serve as plug-and-play chips. Kumar said that it is creating a "firmware to application" software stack for programming the chips.Kumar added: "Customers' workloads can be easily deployed on our more efficient hardware while still using their existing models and databases, bringing them immediate benefits. Currently, Rivos is profitable, and it plans to make money by charging customers (mainly large data center operators) for hardware and supplementary software solutions. Early investor David Goel said that Rivos' 'low-friction' adoption channel is a key differentiator in the competitive chip market.
"The Rivos team has ingeniously integrated the groundbreaking new RISC-V architecture with creative accelerators, effectively turning this vision into reality," Goel told TechCrunch. "Their prototype chips convincingly demonstrate their unique capabilities."
Intense competition
As the generative artificial intelligence boom continues, large tech companies (one of Rivos' potential customer bases) are racing to develop their own in-house chips for artificial intelligence and big data analysis.
Google, on its fifth-generation TPU, recently released Axion, its first dedicated chip for running models. Amazon has several custom chip series. Microsoft launched the Azure Maia AI accelerator and Azure Cobalt 100 CPU last year. Meta is also steadily advancing its own designs.
Meanwhile, dozens of startups are vying for a share of the custom data center chip market, which could reach $10 billion this year and double by 2025.
Groq, a company developing chips that run faster than traditional hardware, recently established a new business unit focused on enterprise applications and use cases. Tenstorrent, an artificial intelligence hardware startup led by engineering celebrity Jim Keller, is seeking to build its chip set into data centers. South Korean fabless artificial intelligence chip company Rebellions has raised hundreds of millions of dollars in funding to increase production of its data center-focused chip Atom.
Currently, Nvidia is the dominant force in the AI chip sector, and it has proven difficult to overthrow.Nvidia briefly became a $2 trillion company this year, thanks to demand for its GPUs used for AI training. Wells Fargo Securities estimates that Nvidia has a 98% market share in the data center GPU sector. The company's data center business grew by over 400% in the fourth quarter of 2023, as Nvidia established a new division to design custom chips for cloud computing and other companies.
Given the intensity of competition, and Nvidia's dominant position which has had a chilling effect on the funding of potential rivals, it has been tough for some upstarts in custom server chips.
A few months ago, it was reported that after a deal with Microsoft fell through, Graphcore's valuation was slashed by $1 billion. The company stated that it was planning layoffs due to the "extremely challenging" macroeconomic environment. Habana Labs, an AI chip company owned by Intel, laid off about 10% of its workforce last year. Similarly, last year, SiFive, like RISC-V startup Rivos, fired 20% of its employees and halted its core product line.
So, will Rivos fare better?
Kumar is reluctant to discuss clients, and Rivos' chips are not expected to achieve mass production until sometime next year. However, Kumar stated that Rivos, with 375 employees and hundreds of millions of dollars in the bank, is well-positioned to scale up manufacturing and double down on platform and software engineering efforts.
Kumar said: "The rapid evolution of generative AI and the convergence with the data analytics stack make it crucial for accelerators to be easily programmable and debuggable, and for data to move seamlessly between the CPU and accelerator. Rivos meets this need with our 'recompile instead of redesign' approach."
Comment