©2026 Business Next Media Corp. All Rights Reserved. No.102, Guangfu S. Rd., Da'an Dist., Taipei City 106, Taiwan
Ambarella CEO: Despite 99% of Funds in Cloud, Edge AI Will Win Half the Market
The 2025 North America Taiwan Tech Summit opened on September 20 (U.S. time), this year centering on the theme “Connect, Innovate, Invest” and focusing on seven major areas of innovation—including AI chips, generative AI applications, and robotics. The event kicked off with an opening address by Fermi Wang, co-founder and CEO of Ambarella, a leading U.S.-based AI chip company.
Ambarella is a key player in edge AI and computer vision, specializing in high-performance system-on-chips (SoCs) used for advanced image processing and deep neural network computing. As a serial entrepreneur, Fermi Wang has weathered the dot-com bubble, the 2008 financial crisis, and the recent U.S.-China trade war, all while staying committed to AI and facing risk head-on.
Fermi Wang notes that although funding is still largely concentrated in the cloud, the growing need for real-time processing, low costs, and privacy will inevitably usher in the era of edge AI. In his view, edge AI will eventually stand on equal footing with cloud AI, each claiming half the market.
How Fermi Wang Navigated Risk and Achieved Startup Success
“Most people only pay attention to success stories, but those stories are usually sugarcoated,” says Fermi Wang. He believes that while entrepreneurs need to talk about their vision, they should pay even more attention to the risks and challenges along the way.
A graduate of National Taiwan University’s Department of Electrical Engineering and holder of a PhD from Columbia University, this serial entrepreneur shared three pivotal moments of risk management from his own career.
The first major challenge came around the year 2000, when Fermi Wang’s multi-core CPU startup, Afara Systems, despite its breakthrough technology, was hit hard by the dot-com crash and the capital drought following 9/11. Fermi Wang recalls that all of Silicon Valley was starved for funding, with almost no investment for two years. In the end, Afara ran out of cash and was forced to sell to Sun Microsystems. This experience taught him a vital lesson: cash flow is the lifeblood of any startup.
This lesson deeply influenced the way he managed finances when founding Ambarella in 2004. In the company’s first 18 months, Fermi Wang completed three funding rounds, ensuring Ambarella had plenty of cash reserves. So, when the 2008 global financial crisis hit and many companies were forced to cut staff and scale back, Ambarella instead expanded hiring and invested heavily in R&D. This laid a solid foundation for the company’s successful 2012 listing on NASDAQ.
“Even in the good times, you have to prepare for winter,” Fermi Wang stresses. Entrepreneurs should plan for the worst, he says, especially during today’s AI boom. It is crucial to consider the possibility of a bubble bursting and to prepare in advance for a cash crunch.
The second major blow came in 2018, during the U.S.-China trade war. At the time, Ambarella’s three biggest customers, Hikvision, Dahua, and DJI, were all placed on the U.S. entity list. Overnight, nearly 45% of the company’s total revenue was at risk of vanishing. Under enormous pressure, Fermi Wang made a bold decision to double down on investing in the future.
Ambarella had already anticipated the AI trend and started developing edge AI solutions back in 2013. By 2018, the team recognized the importance of Transformers, the foundational technology for generative AI, and decided to make it the core of their third-generation AI chips. Fermi Wang convinced the board that even if revenue declined, the only way forward was to create new technology and build a new customer base. Since 2018, Ambarella has invested more than $1.2 billion in R&D. This bold bet ultimately led the company to a successful transformation and record-high revenues.
From Autonomous Vehicles to Drones, Robotics Is Set to Dominate the Edge AI Market
Speaking about the generative AI wave triggered by ChatGPT, Fermi Wang admits that this transformation has been a shock for every company.
Although 99% of investment currently goes into the cloud, Fermi Wang is convinced that edge AI will be the next main battleground, ultimately reaching a fifty-fifty split with cloud AI. He explains that edge computing has irreplaceable advantages in three areas: real-time response, such as autonomous driving, cost, by lowering data transmission and storage expenses, and data privacy.
Fermi Wang predicts that robotics will be the largest market for edge AI, with self-driving cars and drones representing just specialized forms of robots. In the future, all robots will be directly controlled by a large AI model, with sensor input fed straight to the model to drive robot behavior. This places major demands on chip architecture. The key technical challenge is how to run complex, large-scale models under limited power consumption and passive cooling.
As system integrators like Tesla move into chip design, chip companies like NVIDIA build out system platforms, and software giants like Google compete on every front, the competitive landscape is becoming chaotic. Fermi Wang believes the opportunity for startups lies in defining the most efficient solutions. For Ambarella, this means moving beyond being a pure chip vendor and evolving into a true system solutions company.
“We are no longer just selling products. We are trying to show our customers how to run large language models as efficiently as possible on our chips, helping them get to market faster and more easily,” says Fermi Wang.
Fermi Wang also offers advice for the future of Taiwan’s startups and talent. He points out that Taiwan’s talent market is now as competitive as Silicon Valley, and companies must take a global approach, just as Ambarella did when acquiring an autonomous vehicle software team in Italy, searching worldwide for top talent. He also calls for Taiwan to develop a more friendly and modern immigration visa system to attract global talent and truly become the next Silicon Valley.
In an era where AI technology is evolving at ten times the usual pace, how should entrepreneurs respond? Fermi Wang’s answer is to go back to basics. While the application layer may change rapidly, the fundamental disciplines—science, physics, mathematics—and the critical thinking and foundational skills developed through education, will always be essential and irreplaceable.
Original article: 99%投資流向雲端AI，安霸執行長王奉民：最終邊緣AI仍能占有半壁江山
©2026 Business Next Media Corp. All Rights Reserved. No.102, Guangfu S. Rd., Da'an Dist., Taipei City 106, Taiwan