“
This is a record written by someone present—a RISC-V idealist, traversing the path between two startups. He once illuminated the free fantasies of a group of people with a few development boards, and after finishing a cigarette on the balcony late at night, silently folded the group.
We are not just talking about him. We are discussing the obsession, struggles, debugging, sleepless nights, and relentless efforts of an era, along with the few heated silicon chips left behind and a line stating “booting Linux on RISC-V.”
That line made him believe he could change something.
In the autumn of 2025, Shenzhen is still hot and humid, and the streets of Xixiang are still damp. In a regular two-bedroom apartment, Ke Yiran sits in front of his computer, and the content on the screen is no longer the familiar Linux kernel logs, but a brand new AI model training dashboard. Two dogs breathe evenly at his feet, and his girlfriend is hanging clothes on the balcony, with sunlight slanting onto her rolled-up sleeves.
The electric scooters downstairs still roar, and life seems unchanged. But in his heart, something has completely come to an end—the dream of faith in RISC-V has finally dispersed.
1. Youthful Ideals and the Faith in Silicon
Ke Yiran was born in the early 2000s in Chaozhou, where his family ran a small business, making them an “atypical Chaoshan family”—he is the only child.
“My dad’s generation talks about inheritance; having many brothers means having power. But my mom was not in good health, so I ended up being the only one.” He said this in a flat tone, without sadness, as if narrating a product lifecycle.
Being an only child allowed him to engage with the electronic world early on. He enjoyed dismantling radios and modifying sockets in elementary school, and burning out power boards was a common occurrence. Having been electrocuted more than once, he recalls with a smile: “That burnt smell is more fragrant than plastic.”
Before graduating from university, he accidentally entered a small hardware company in Shenzhen for an internship—Wamos Technology, which was then undertaking the design project for the RISC-V development board Vision Five from SiFive China.
This collaboration led Wamos to decide to adopt RISC-V as one of the company’s future development directions, quickly incubating a new brand and subsidiary: Milk-V.
As one of the earliest founding team members, Ke Yiran joined the core team of Milk-V when the project was just starting.
At that time, he was in his early twenties, speaking quickly and logically, with a hint of the sharpness of a newcomer—he believed that the order of the chip world was waiting to be rewritten.
2. Milk-V: A Hardware Experiment from Zero to One
By 2022, the Chinese semiconductor industry was experiencing a tumultuous period. Large models had not yet gained popularity, but chip bans had already arrived. Numerous domestic RISC-V manufacturers sensed an opportunity: companies like T-head, SiFive, Xinlai, and Jingxin began launching SoCs, and toolchains and software stacks started to climb.
Ke Yiran transitioned to the newly established Milk-V company with Wamos, fully responsible for building the product market and developer ecosystem.
Internally, he was the least managerial VP—spending every day with boards, checking logs, writing bug reports, and arguing with the supply chain over BOM prices.
“At that time, we were truly starting from scratch; many chips didn’t have datasheets, and many drivers were modified from chip test codes.”
The Duo, Mars, and Pioneer development boards were mass-produced within a year. During that time, he hardly had any weekends.
Pioneer was not an ordinary development board—it was Milk-V’s ambitious project, a 64-core RISC-V server motherboard aimed at small-scale servers and cloud development. The team regarded it as a stepping stone to prove that “RISC-V can be used in servers,” choosing to launch a crowdfunding campaign on CrowdSupply, with pricing strategies, early supporter packages, and community target pages all personally handled by Ke Yiran. He wrote copy, filmed demo videos, ran community AMAs, and stayed up late modifying the FAQs on the crowdfunding page. For those months, he became the person in the company most like a “vote collector.”
More dramatically, Pioneer was sent to Linus Tech Tips for review—this was a significant emotional leap for Ke Yiran. He had been watching Linus’s channel daily since high school; Linus’s unpolished jokes and obsession with hardware had inspired him to secretly watch videos while coding late at night. Seeing his own board being dissected and explained by a guru, with the camera zooming in on the PCB silkscreen, brought him an inexplicable sense of moisture: childhood admiration transformed into real hardware exposed to the world. At that moment, he felt like a child who had completed a rite of passage, forgetting his fatigue, leaving only pride.
The video garnered over 1.3 million views, and many foreign developers heard about RISC-V motherboards for the first time, making Pioneer one of the few domestic development boards featured on mainstream tech channels.
Pioneer’s crowdfunding ultimately raised over $250,000, a rare achievement among open-source hardware projects that year, significantly boosting the team’s confidence. For Ke Yiran, this not only validated the popularity of RISC-V within the developer community but also proved that his obsession with “allowing more people to touch RISC-V” was not a fantasy.
“I remember once debugging the HDMI driver, running EDID all night, and burning three boards.” The next day, he attended the meeting as usual, still smelling of burnt components. He smiled, his eyes showing fatigue, “This industry is like that; burning boards is also part of iteration.”
Their goal was never grand—just to let more people “touch” RISC-V.
“It’s like when you first light up a development board that can burn its own system; even if the desktop lags and the drivers are incomplete, you know that the fate of this machine is no longer controlled by others.”
3. The Rise of Ideals: From Boards to Ecosystems
In the spring of 2023, Duo began mass production, and within just three months, sales exceeded ten thousand units. At that time, this was almost a miracle among domestic RISC-V development boards.
He wrote on X:
“
“This is not just a board; it is a signal. The spring of open-source hardware is coming.”
That year, the RISC-V community in China was as lively as if it had just received a shot of adrenaline. Events like the Xuantie RISC-V ecosystem gathering, the RISC-V China Summit, and the establishment of the RISC-V Working Committee saw people gathering around Milk-V boards for photos at every exhibition.
Ke Yiran began to frequently appear at various community events, bringing boards, prototypes, and a fervent technical faith. He remotely debugged SPI drivers with foreign developers on Discord, replied to kernel issues on GitHub in broken English, and even wrote cross-compilation tutorials, explaining them as simply as making a plate of Chaoshan oyster omelet.
He believed that as long as RISC-V could run Chromium, VS Code, and Ollama, the revolution in desktop computing would begin at that moment.
His goal was never about shipment volume, but about “bringing freedom to every device that can be powered on.”
4. The Cracks of Reality: Upstream Dilemmas
However, starting in 2024, this “freedom” project began to lose momentum.
The RISC-V community was not built like ARM or x86, with unified standards and commercial driving forces. Chromium’s CI has always been inadequate on RISC-V, WebAssembly support crawled like a snail; LLVM’s new instruction optimizations were delayed in merging, and GCC maintainers were stretched thin; each chip manufacturer had custom bootloaders, fragmented drivers, and programming tools that were incompatible with each other.
“You fix the audio, and the USB goes down; you just got SPI working, and HDMI goes black again.” He said, “It’s like playing whack-a-mole; you can never finish fixing everything.”
More realistically, the company began to shift towards “finding a commercial model to make up for losses.” Milk-V’s revenue growth slowed, and hardware profits were continuously squeezed by chip and logistics costs. Investors began to ask, “Can we add AI? Can we ride the wave of large models?”
And when he flipped to the tenth page of the PPT, he suddenly fell silent.
“They want results; I care about the process. But the process… sometimes it can starve you.”
5. The Choice Between a Chaoshan Father and Shenzhen
During that time, his father came from Shantou to visit him in Shenzhen.
The atmosphere at the dinner table was tense, and the logic of the old businessman was suffocating:
“Stop messing around. Come back, and I’ll start a company for you, and you can have a car too. Don’t you like electric cars? I’ll get you a SU7 Ultra; you can have whatever you want.”
Ke Yiran shook his head, “Dad, I’m staying in Shenzhen.”
At that moment, he realized he had severed ties with the path of Chaoshan.
He rented a small apartment with a balcony, living with his girlfriend and two dogs. Every night after walking the dogs, he would open his laptop and watch lines of compilation logs scroll by; that was his way of combating anxiety.
6. The Night of Disillusionment
Just after the Spring Festival of 2025, the enthusiasm for RISC-V visibly declined.
The topics in the community shifted from “next-generation SoCs” to “how to be compatible with ARM”; originally active developers began to turn to other projects, and only the mailing list continued to operate on inertia. Milk-V began to frequently hold “strategic adjustment meetings,” with terms like layoffs, project pauses, and funding obstacles appearing repeatedly in weekly reports.
“Are we still going to make the next board?” a colleague asked.
He did not answer.
That night, he returned home alone, lit a cigarette on the balcony, and gazed at the skyline towards Nanshan. The dog lay at the door, and the wind was gentle.
He suddenly recalled the drivers he had written, the patches he had pushed, the NDAs he had signed, and the late nights arguing with factories over a few cents.
The code is still there, the boards are still there, but that dream has dissipated.
“It’s not that the dream shattered; it’s that I finally woke up.”
7. Intel Stocks and New Ironies
Fate always loves irony.
A few months later, Intel’s stock price soared after NVIDIA suddenly announced an investment and reached a cooperation agreement with it. The options he bought at a low point a few years ago quadrupled.
That investment was originally just his alternative bet on the RISC-V community.
“At that time, I thought, since Intel is investing in RISC-V, I might as well buy some Intel stocks as a way to support open source.”
Little did he know, this became his startup capital for leaving the RISC-V scene.
He resigned from his VP position and registered a new AI startup, focusing on “voice-driven AI toys”—targeting scenarios like parent-child interaction, companionship, and storytelling, using generative voice models to enhance human-computer interaction experiences.
A friend joked, “Does this count as… using Intel’s money to redeem yourself from RISC-V?”
He laughed, “Yeah, it’s a bit ironic. But ultimately, you have to survive.”
8. From Chips to Models: A Shift in Ideals
The new office is still in Baoan, just a few hundred meters from the small rented room where he debugged Pioneer.
This time, he no longer writes DTS, no longer debugs serial ports, and no longer has his nostrils blackened by burnt boards.
In his meeting room, discussions revolve around cleaning voice datasets, token control, and anti-mis-touch logic for children’s scenarios; the team’s daily meetings are no longer about “U-Boot not starting,” but “this round of distillation lost 1.7 points in accuracy.”
His new company’s slogan is not “Let architecture be free” but:
“
“Make companionship smarter.”
This phrase feels like a farewell letter to his past self.
He still keeps a few old boards. Duo, Mars, Pioneer, labeled with white tags, sit silently on the shelf like collectibles.
Those were the brightest pieces of silicon in his life.
9. Epilogue: The Ashes of Faith
In the quiet of the night, he sometimes still opens the RISC-V discussion group that has been silent for a long time, now pushed up by new messages, watching the sparse discussions about which new RISC-V chip corresponds to the performance of which ARM chip.
He no longer participates, does not discuss, nor leaves the group.
He simply silently turns on Do Not Disturb and folds the group away.
In a corner of his old computer, the familiar boot log is still saved:
“
Booting Linux on RISC-V… init: starting pid 1, tty ‘’: ‘/etc/init.d/rcS’
He renamed the file:
dream.log