MIPS is a historic CPU IP core company with a tumultuous business and management history. Under the leadership of new CEO Sameer Wasson, formerly of TI, MIPS is writing a new chapter. Can he resurrect MIPS from obscurity and achieve solid growth? Wasson’s primary task is to explain the current reality of MIPS.Three months after joining MIPS, Wasson said, “We have just begun. We have a lot of work to do.” MIPS plans to increase its workforce from 62 employees when Wasson joined to around 100 by the end of the year.Recently, rumors of layoffs at SiFive have become a source of talent for MIPS. Wasson stated, “We have indeed hired some of those who were laid off. But we are also recruiting more who were not laid off.”However, the specific personnel list will be announced at CES in January.
MIPS’s Advantages
Under the leadership of the previous management, MIPS underwent a difficult transition to RISC-V. A year ago, the company launched the P8700, the first licensable CPU using the RISC-V instruction set. Reportedly, the P8700 combines a proven MIPS-compatible core and outperforms most other RISC-V designs.Key player in the ADAS market, Mobileye, is already a customer of the P8700. Earlier this year, the company announced its processing core would switch from MIPS to RISC-V.A few weeks ago, MIPS also signed an agreement with an undisclosed customer in the data center market. Wasson indicated that this customer is “one of three hyperscale computing companies.”Wasson’s primary task is to narrate the story of the reborn MIPS: the company’s current status, its strategy regarding RISC-V cores and MIPS instruction set architecture, how it differentiates itself in the RISC-V community, and where it plans to focus its business.MIPS’s LegacyTwenty years ago, MIPS and Arm competed fiercely in markets such as smartphones. Clearly, MIPS lost that battle. However, since then, Wasson pointed out, “MIPS has created a very good niche market in specific areas,” and he is determined to leverage that.He referred to MIPS’s advantages in networking, which is closely related to data transmission. MIPS addresses issues through efficient network connectivity and multithreading execution.Wasson noted that a major issue in today’s AI field is “efficient data transmission,” and “we can address data transmission issues more efficiently by using the MIPS architecture on top of RISC-V. We can bring value to customers who lack this capability.”However, Wasson emphasized that MIPS does not aim to challenge Arm. “I do not intend to turn MIPS into something it is not.” It is impossible to have wild dreams of challenging Arm in the server market or competing with Tenstorrent to build the next Nvidia. “We will not do that.”Instead, Wasson focuses on MIPS’s recognized strengths: “Efficiently transmitting data better than any other engine on Earth.” He stated, “We will do this in a deterministic, low-latency manner. This is something others cannot achieve.”Wasson indicated that MIPS will pursue specific markets compatible with MIPS rather than trying to be a jack-of-all-trades that confronts everyone and everything head-on. This approach may have led to the mistakes made by SiFive.
Automotive Sector Layout
Given that Mobileye “completely relies on MIPS,” Wasson stated that MIPS can claim to occupy 60-70% of the ADAS market share. He added that MIPS’s multithreading capabilities are particularly prominent when multiple sensor inputs process data in the ADAS system.
Software-defined vehicles are another automotive segment where Wasson believes MIPS can thrive. “If you consider the zonal software-defined vehicle domain, which builds multiple environments on the same chip, MIPS can enable you to achieve hardware virtualization.” This offers significant advantages compared to cycle-consuming software virtualization.
Wasson’s strategy is to first identify the total addressable market, then more closely determine in which product categories MIPS has architectural advantages. “Then, we can implement these advantages on our RISC-V, allowing customers the freedom to include instructions.”
Today’s MIPS is RISC-V
The name MIPS in the industry signifies the MIPS instruction set architecture (ISA) and MIPS cores. Wasson acknowledged, “This is somewhat confusing. But today, every new development we undertake at MIPS is based on RISC-V, and we are no longer developing MIPS. We support the MIPS ISA, but every new core we create is based on RISC-V.”
In short, Wasson stated, “The MIPS ISA is no longer applicable for any new designs.”
Wasson explained that MIPS spent three years converting the MIPS ISA into an optimized RISC-V implementation, “while maintaining the same architectural elements that made MIPS successful.” MIPS’s cores, including multithreading and hardware virtualization, still exist in the architecture, “we have not lost those.”
MIPS’s Revenue
So, how is the revenue distribution between the old MIPS and RISC-V?
Wasson stated that it is currently about 50:50.
For MIPS, the royalty revenue from the old MIPS allows the company to continue operating without leaving investors high and dry. But Wasson said, “In the past 12 months, 50% of our revenue has come from new licenses.” Of the new licenses, half comes from existing customers re-signing with RISC-V, and the other half from new customers.
Many RISC-V companies are private. Wasson stated, “But as far as I know today, when it comes to revenue, I believe MIPS is absolutely in the top two or three.”
To distinguish itself from other RISC-V companies, MIPS needs to continue investing, hiring top talent, and focusing on specific market segments. MIPS’s plans for targeted segments in data centers, automotive, and embedded markets remain undisclosed. However, Wasson indicated that strategic plans have been drafted and will soon be shared with the team.
When asked how to describe MIPS’s model in the RISC community, Wasson emphasized the importance of focus. He said, “We are not trying to enter the ultra-low-end market that Andes (another RISC-V competitor) is pursuing. But this is a very smart strategy, and Andes has been very successful in focus and execution.”
He added, “So think of it this way… at MIPS, we are doing what Andes is doing, but targeting high-performance or specific markets that differ from those pursued by Andes.”
From TI to MIPS
When MIPS invited Wasson to consider the CEO position, his decision was not easy. After 18 years at TI, he had become “very comfortable.” As Vice President of Processors and Business Unit Manager, he had established TI as a mainstream supplier of Arm-based MCUs for automotive and industrial markets.
But he noted, “As Arm continues to ‘simplify’ things for everyone, our frustration has been growing, as we feel increasingly squeezed and find it harder to differentiate our products from others.”
Arm’s advantage lies in its ability to “get you on board… first the core, then the interconnect, then the NPU, and then software certification.” After that, everything is Arm’s, or nothing. Because once you are on board, it is difficult to get off, and harder to get back on.
He said, “Around that time, TI began to notice RISC-V.” Wasson realized that RISC-V was about “regaining control.” Process node improvements were slowing, and the benefits of finer nodes were diminishing. Arm was taking over the technology roadmap. Wasson had to ask at TI: “How will we spend R&D funds and bring value to customers, investors, shareholders, and everyone?”
Back to the Whiteboard
Wasson said, “I also felt it was time to go back to the whiteboard from an architectural perspective.”
Wasson observed, “When you start innovating, you realize maybe we need to address computing problems differently and regain flexibility in architecture. Frankly, that’s why RISC-V is so interesting.”
Wasson hesitated for five months before leaving TI.
But it turned out that MIPS’s fierce competition for a CEO lasted over six months. During this time, the board recognized the value of MIPS and was willing to reinvest cash flow back into the business.
Wasson stood out among more than 30 candidates, believing he had the board’s trust. “I have operational freedom, I have cash flow… I have a market there… I can make an impact.”
Wasson said it was time to decide whether to fish or cut bait, “If I can’t execute, I have no one to blame.”
The MIPS computer system was founded 40 years ago by John Hennessy and Chris Rowen, aiming to commercialize Stanford’s MIPS architecture, a groundbreaking RISC design. Since then, MIPS has changed hands multiple times, belonging to Silicon Graphics in 1993, Imagination Technologies in 2013, Tallwood VC in 2017, and Wave Computing in 2018. Although Wave sought bankruptcy protection in 2020, MIPS thrived, and Wave became MIPS.
MIPS has found the secret to retaining the advantages of the MIPS ISA while allowing people to develop extensions and differentiated products based on RISC-V. Theoretically, Arm could also learn from this strategy in the future. “Nothing can stop them from doing this,” Wasson said, “but it will be a long journey because they have a very strong tradition of architecture licensing.”
from A to B
