Axera's Qiu Xiaoshen: High-Computing Chips to Become Major Growth Engine Next Year

Domestic Chips Enter a Window of Opportunity for Automotive Integration

Against the backdrop of the automotive industry's continued emphasis on cost reduction and efficiency improvement, automakers' supply chain choices are undergoing profound changes. The automotive chip market, once firmly controlled by international giants, is gradually tilting toward more cost-effective domestic solutions. As a key player in China's AI chip sector, Axera has keenly identified this structural opportunity and is accelerating its push into the high-computing assisted driving chip space.

On the eve of the Beijing Auto Show, Axera joined forces with Qianli Technology, StepStar, and other companies to establish the "Qianli Alliance," aimed at building an open industry ecosystem around intelligent driving chips. At the same time, Axera announced it would officially launch its high-computing assisted driving chip M97 in Q3 of this year, mounting a charge into the advanced intelligent driving market.

Qiu Xiaoshen: M97 Successfully Taped Out, Set for Mass Production Vehicle Integration in 2026

Axera founder and CEO Qiu Xiaoshen stated publicly at the Beijing Auto Show: "The high-computing product has been successfully taped out and is entering its normal development cycle." According to the roadmap, the M97 will be gradually integrated into multiple mass-production vehicle models starting in 2026, becoming a critical engine for the company's performance growth.

In terms of revenue structure, Axera currently remains dominated by traditional terminal computing products, which contribute over 80% of the company's revenue, while automotive and edge computing business revenue accounts for a relatively small share. However, Qiu Xiaoshen is confident about future shifts in the business landscape. She expects that over the next three years, as high-computing products like the M97 progressively enter mass production in vehicles, the automotive business will significantly increase its share of overall revenue, fundamentally rewriting the current revenue structure.

This means the period from 2025 to 2028 will be a critical window for Axera's transformation from a traditional terminal AI chip company to an automotive high-computing chip company.

High Bandwidth + Low Power Consumption: M97's Core Competitiveness

When discussing the M97's technical advantages, Qiu Xiaoshen pointed directly at current industry pain points. She stated: "Current mainstream advanced assisted driving chips have a fairly significant problem — insufficient bandwidth. After recognizing this issue, we substantially increased bandwidth in the M97's design."

Qiu Xiaoshen further explained the technical logic behind effective computing power. The effective computing power of an assisted driving chip is determined by two key factors: the number of computing units and bandwidth, meaning the speed of data read and write operations. She emphasized: "Even with 2,000 TOPS of computing power, if DDR bandwidth is insufficient, the chip cannot deliver its full 2,000 TOPS capability. A major advantage of our chip is its very high DDR bandwidth, which allows computing power to be fully utilized. At the same time, our chip has excellent power consumption control, which is another advantage."

In other words, the M97's design philosophy is not about blindly stacking computing power numbers but rather solving the problem of "computing power utilization" at the system level. At a time when the industry generally pursues high TOPS figures, Axera has chosen to establish differentiated barriers along the two dimensions of bandwidth and power consumption — a notably forward-looking approach. To maximize the chip's actual computing power, Axera has conducted deep optimizations in architecture design, striving to ensure every computing unit operates efficiently.

Industry Ecosystem Strategy: The Strategic Significance of the "Qianli Alliance"

Notably, Axera has not chosen to go it alone but is instead building a collaborative ecosystem through the establishment of the "Qianli Alliance." Alliance members include intelligent driving solution provider Qianli Technology and large language model company StepStar, among others. This combination covers the complete chain from chip-level infrastructure to algorithm applications and reflects Axera's forward-looking positioning amid the trend of on-device AI and large model convergence.

As advanced intelligent driving gradually penetrates from high-end to mid- and low-end vehicle segments, market demand for cost-effective high-computing chips will see explosive growth. If domestic chip manufacturers can find the optimal balance among performance, power consumption, and cost, they stand to win considerable market share in this round of industry reshuffling.

Outlook: The Path to Breakthroughs for Domestic Intelligent Driving Chips

From a broader perspective, Axera's M97 is not just a product — it represents a signal that domestic intelligent driving chips are charging into the advanced market. Over the past few years, NVIDIA's Orin series has nearly monopolized the domestic advanced intelligent driving chip market, while domestic solutions such as Horizon Robotics' Journey 6 and Black Sesame Technologies' Huashan series have been striving to catch up. With its differentiated advantages in bandwidth and power consumption, Axera is well positioned to carve out its own ecological niche in the competition.

Qiu Xiaoshen has a clear plan for the company's development path over the next three years: complete M97 customer onboarding and adaptation in 2025, achieve mass production integration in multiple vehicle models in 2026, and drive the automotive business to become the company's core revenue pillar from 2027 to 2028. If this roadmap is successfully executed, Axera will complete a critical leap from an emerging edge AI chip player to a major force in automotive high-computing chips.

Driven by the dual forces of domestic substitution trends and accelerating intelligent driving adoption, the competition in high-computing chips has only just begun. Whoever can first establish advantages simultaneously across performance, cost, and ecosystem stands to become the dominant force in the next phase of the intelligent driving chip market.