miHoYo has invested in a prominent female entrepreneur
Wealthy and passionate miHoYo has made another investment.
This week, Ninejoy Semiconductor completed a new round of financing. Founded by Liu Hongjie, a female engineering PhD, this company focused on SRAM computing-in-memory chips has raised a total of 1 billion yuan in accumulated financing to date, with miHoYo prominently listed among the investors in its latest financing round.
Coincidentally, miHoYo also recently participated in the Series A financing of another computing-in-memory chip firm, MindChip Prime.
It is still fresh in memory that half a year ago, Liu Wei, known as "Da Wei Ge" at miHoYo, announced that the company would invest up to 1 trillion yuan in large models over the next three years, and later set a clear goal: "Within 2 to 3 years, miHoYo will become a pivotal player among domestic large model teams."
miHoYo's grand AI ambition is steadily being turned into action.
Shenzhen, Returned Female PhD, Backed by miHoYo
Looking back at the ups and downs of semiconductor investment over the past decade:
From 2017 to 2019, shifts in the international landscape and the launch of the STAR Market ushered in a boom for the industry, followed by several years of red-hot financing activity; the market then suddenly braked sharply in 2022, and did not start to rebound and build momentum until 2024.
As a result, 2018 was a fairly active year. At 30 years old, Liu Hongjie chose to settle in Shenzhen to launch her startup, targeting the new-generation computing-in-memory architecture to explore innovative chip design.
It is quite rare that Liu Hongjie is one of the few female entrepreneurs in China's semiconductor industry.
Liu Hongjie
Born in Jining, Shandong Province, she completed her bachelor's degree in Electronic Science and Technology at Jilin University, then went to Nanyang Technological University in Singapore for further study, obtaining a master's degree in Integrated Circuit Design. She later worked at a leading tech enterprise, before deciding to pursue her PhD at ETH Zurich in 2015.
In 2016, before graduating, she returned to China and took part in the China Innovation and Entrepreneurship Competition in Shenzhen, winning first place in the team category of the electronic information track. As her academic achievements and research direction attracted widespread attention, investors persuaded her to start her own business at that time, sowing the first seed of her venture.
"Chip development requires defining future products three to five years in advance," Liu Hongjie once noted in a public interview. At the time Ninejoy Semiconductor was founded, most industry investors favored the mature GPU and CUDA ecosystem, and held doubts about the real commercial landing of computing-in-memory technologies such as SRAM.
On this path where no industry consensus had been reached temporarily, Ninejoy Semiconductor kept tackling technical bottlenecks and built a multi-level system "centered on computing-in-memory, supplemented by near-memory computing". It has now completed the R&D of multiple generations of computing-in-memory chips, with its product portfolio covering HBF SSD, ADA300 series computing-in-memory chips, ADA400 inference boards and AI inference servers.
A review of its financing history shows that Ninejoy Semiconductor went through a three-year long cold spell for financing, but eventually gathered a lineup of well-known investors from market-oriented, industrial and state-owned capital backgrounds one after another:
In 2018, it closed its angel round with participation from Hongtai Fund and Lanpu Venture Capital; in 2019, MiraclePlus invested in its pre-A round; in 2021, Pudong Tech Innovation and Weihao Core Chip led its Series A round, with Haiwang Capital, Huayi Venture Capital and other institutions participating. Three years later, in 2025, Yuanhe Puhua led its Series B round, joined by Woniu Capital, Ningbo Zhenhai Hanyun Equity Investment, Gaoshan Xinyu, Yinghao Capital and other investors.
In 2026, its financing pace suddenly accelerated, with three consecutive rounds completed, including the B+ round in January, the B++ round in April, and this latest round in October. According to the disclosure from Yiwei Capital, Meituan Strategic Investment, XPeng Motors (Xinghang Capital), Longsys Industrial Fund, Yuanhe Puhua, OmniVision Group and other strategic investment institutions have entered the market successively, and the latest new participants are Zhiwei Capital and miHoYo Strategic Investment.
At this point, storage manufacturers, game companies, mobile phone vendors, automakers and leading internet giants have all gathered in the sector.
Two Investments in a Single Week, The Track Begins to Boom
In just one week, miHoYo has made two investments in chip enterprises.
On October 19, Beijing-based computing-cooperative chip firm MindChip Prime also announced the completion of a Series A financing of over 400 million yuan, with Boyuan Capital, miHoYo, GigaDevice and other parties listed among its investors. Similarly, Academician Liu Ming, the initiator of MindChip Prime, is also an outstanding female figure in the industry.
This reminds people of the previous report from *The Information* that miHoYo plans to invest up to 1 trillion yuan in the artificial intelligence R&D sector over the next three years, and is currently sourcing chips to train more competitive AI models.
However, these two investments by miHoYo are not targeted at GPUs, but at two enterprises in the computing-in-memory track.
The concept of computing-in-memory can be explained with a simple analogy: if your home is extremely far from your workplace, you have already used up 80% of your energy before you even arrive at your workstation — this is exactly the problem plaguing traditional chips, where data needs to travel back and forth frequently between storage and computing units, wasting 80% of power consumption on the transmission path. When undertaking large model inference tasks, such data movement costs will be directly passed on to each Token, leading to higher power consumption and operating cost issues.
In this context, computing-in-memory chips are designed to eliminate unnecessary data movement, so that computing operations can happen directly inside the storage units. In the AI large model era that prioritizes efficiency, this technology has become increasingly critical, and can even be regarded as one of the biggest bottlenecks restricting the release of AI computing power. For miHoYo, betting on this direction means betting on the structural demand for underlying computing power brought by future AI games.
The timing of this bet is also very opportune.
The ongoing AI boom has pushed the industrialization of computing-in-memory chips to a critical turning point, with supporting guidance also released at the policy level: the "15th Five-Year Plan" issued by the Ministry of Industry and Information Technology and the National Development and Reform Commission this September has included computing-in-memory as a key R&D priority for end-side intelligent computing. In addition, Beijing's digital economy plan also lists computing-in-memory chips and advanced packaging as cutting-edge R&D priorities.
In terms of financing performance, 2026 is undoubtedly a year when the financing pace of the computing-in-memory track has accelerated sharply, with large-scale financing deals emerging almost every month:
In October, Suanmiao Technology completed its Pre-A2 round and Series A financing, with total financing reaching nearly 2 billion yuan within the year. In the same month, Siminwei secured a nearly 100 million yuan seed round financing and a 100 million-yuan-level angel round financing. In September, 3D computing-in-memory AI chip firm Micro-Nano Core announced the completion of a 1 billion yuan Series C1 financing, only three months after it announced the completion of the Series B3 and B4 consecutive financing rounds in June. In mid-August, Qianhe Yibang, a 3D computing-in-memory chip firm established just over two years ago, completed a Series B financing of over 2 billion yuan.
The long-awaited industry boom has arrived.
Trillion-yuan AI Ambition
"Even if we fail, we will take it, it's just like setting off a grand firework", miHoYo is embracing AI with a posture that leaves no retreat.
The public first perceived such determination in May this year, when Liu Wei, co-founder of miHoYo, mentioned at a technology sharing session that the upper limit of miHoYo's investment in large models over the next 3 years is 1 trillion yuan. After that, every time Liu Wei appears in public, AI is always the core topic he emphasizes.
His latest public statement was delivered at miHoYo's 2027 campus recruitment special session at Shanghai Jiao Tong University, where Liu Wei revealed: "Especially after we started developing our Coding model this year, I think our progress is advancing by leaps and bounds. I have strong confidence that within 2 to 3 years, miHoYo will become a pivotal member among domestic large model teams."
Back to the 1 trillion yuan AI investment plan, this figure roughly exceeds miHoYo's total annual revenue. The public is curious about how miHoYo will allocate this 1 trillion yuan, and what kind of AI achievements this exceptionally talented game company will deliver.
As is known to all, miHoYo has already built a huge investment ecosystem over the years, covering many tracks including brain-computer interface, large models, embodied intelligence, nuclear fusion, and aerospace rockets, with star portfolio companies including MiniMax, Geespace, Orienspace and Energy Singularity.
Interestingly, Liu Wei once attributed many of his early investments to a kind of "unintentional yet rewarding" personal sentiment — this April, he reiterated during the on-site Q&A session at Shanghai Jiao Tong University: "I do not engage in professional investment, my investment behavior is purely driven by personal sentiment. Some of the investments are made because I met outstanding talents and hope to support them, and they are not operated in accordance with conventional investment logic."
For example, miHoYo invested in civilian nuclear fusion firm Energy Singularity because the company's founder was a tutor from Shanghai Jiao Tong University who supported miHoYo's entrepreneurship in its early days. Similarly, the brain-computer interface project Zero One Thinking that miHoYo just successfully exited from was also invested because the company's founder was the master thesis supervisor of miHoYo's co-founder Cai Haoyu.
But now, compared with some early cross-border investments made out of personal sentiment, miHoYo is obviously further strengthening its independent investment in AI technologies, just like its recent initiative to invest in the two aforementioned chip companies.
The AI commitment has a three-year timeline. miHoYo has many competitors in the market, but the only real rival that miHoYo needs to surpass is itself.
This article comes from the WeChat official account "Pedaily" (ID: pedaily2012), authored by Feng Yuchen, and published with authorization from 36Kr.