Forest Bai (WhiteForest)

Answering a question about the graduate school entrance exam in 2018, Bai Fengxiao recalled taking the exam in 2014. He said there were many questions he could not answer in the final specialized-subject exam and that he briefly wanted to give up, but he went on to finish the parts he could answer and ultimately made it into the second-round interview as the last-ranked qualifying candidate. This is his own recollection of an earlier exam; the “2019 graduate entrance exam” in the question’s title is not the exam year he described.⁠[1]

In another answer in 2019, he explained that one motivation for taking the graduate entrance exam was to change the career path he then envisioned, of joining a power grid or power generation company after graduation. He saw exam preparation as a period with clear goals in which he could feel daily progress, and believed the experience broadened his life choices. The power companies mentioned were his anticipated destination at the time, not employers he had worked for.⁠[2]

A May 29, 2015 report on the North China Electric Power University (NCEPU) news site recorded that the Baoding campus had held an experience-sharing session for outstanding graduates of the class of 2015 the previous day, at which Bai Fengxiao of the Department of Power Engineering, as an invited representative, shared his graduate entrance exam experience. The report stated explicitly that he had been admitted as a graduate student to the Department of Thermal Engineering at Tsinghua University; his talk covered school selection, self-assessment, study plans, mindset, and interview techniques. This contemporaneous university record is consistent with his later statements that he studied for his bachelor’s degree at NCEPU and his master’s at Tsinghua.⁠[3]

Sponsored content published by Foresight Ventures on The Block stated that in 2017 Forest Bai was pursuing a master’s degree in engineering at Tsinghua University and looking for new entrepreneurial directions between traditional finance and technology problems. The material is paid institutional content.⁠[4]

In a Y TALK interview released in 2026, Forest Bai said he studied in the Department of Thermal Engineering at Tsinghua University, describing his academic background by noting that the department covers engineering fields such as boilers and engines. He later chose a finance direction; this answer adds information about his field of study but gives no new date for the award of his degree.⁠[5]

In June 2021, he further explained on Zhihu that he did his bachelor’s at NCEPU and his master’s at Tsinghua, studying energy and power-related subjects for six years in total, and switched careers upon graduation. He distinguished his choice from that of classmates who continued working in power and engineering design.⁠[6]

In a 2023 interview, Web3Brand said Forest Bai had followed the crypto industry for many years and had moved from a brokerage firm into crypto investment a little over two years earlier. Public secondary sources point to his traditional finance experience as an investment manager at Huarong Securities, but no employment documents give precise dates.⁠[7]

In a 2018 Zhihu answer, he said he had studied engineering for both his bachelor’s and master’s degrees and later moved into the securities industry; before coming to Beijing, he knew very little about the finance industry. He attributed the changes after entering Tsinghua to broader horizons, peers, and career options rather than simply obtaining a degree. The answer provides an early first-person record of his career change before he entered the crypto industry.⁠[8]

In 2020, he also discussed the educational screening he encountered when job hunting: although he already held a master’s degree from Tsinghua, he still encountered positions with requirements on undergraduate institution during campus recruitment and later job changes. He also stressed that graduate study had indeed brought him opportunities; this account is of his personal job-hunting experience.⁠[9]

On June 23, 2019, Bai Fengxiao published “You Complain About Missing Out on Property and Stocks, So Do You Dare to Buy Bitcoin?” on Zhihu, showing that he was already discussing BitcoinBitcoinBitcoin (common symbol BTC) is an open peer-to-peer electronic cash protocol, network, and native digital asset. Full nodes independently verify transactions and blocks according to the consensus rules; transactions are recorded as UTXOs, and blocks are organized into a public ledger through proof of work. Bitcoin is not issued or managed by any single institution, but using, holding in custody, and trading it still involve key, privacy, market, and legal risks.Open the full entry publicly before joining Foresight Ventures. He explained his view of Bitcoin in terms of scarcity, social consensus, and monetary rules independent of the traditional credit system, while acknowledging that price volatility limits its use for payments and as a store of value.⁠[10]

The article also discussed contrarian investing and entry timing. Citing Feng Liu’s idea about divergence between market consensus and common sense, he treated institutional participation, Facebook’s then-announced Libra plan, and changes in the monetary environment as clues in his observation of Bitcoin. The judgments in the article reflect his personal investment views in 2019; the article did not disclose verifiable position sizes or returns.⁠[10]

On risk, he urged readers to watch for concentrated holdings, emotion-driven trading, and noise from opinion leaders, and to be especially wary of other, more speculative tokens. At the end of the article he drew a comparison with the entrepreneurial choices of a classmate from his hometown in Jiangsu, emphasizing the relationship between knowledge, risk awareness, and action. Jiangsu here is the hometown background he described himself.⁠[10]

Multiple contemporaneous and current sources list Forest Bai as a co-founder of Foresight Ventures. A 2022 research report jointly published by BCG, Bitget, and Foresight Ventures directly labeled him as Co-Founder on its authors page; the current official website lists him as Co-founder & Advisor. The firm usually gives 2021 as its formal founding year, while an individual sponsored article gives 2018.⁠[11]

In an August 2021 column on Matrixport’s funding round, Foresight’s own firm profile instead stated that it was founded in 2020 and mentioned a predecessor with ten years of investment experience. This does not match the 2021 and 2018 accounts in other sources, and the available material is not sufficient to clarify the timing relationship between the predecessor and the brand’s formal founding.⁠[12]

In a long post of his own in March 2026, Forest Bai recalled that he entered the crypto industry in 2021, first co-founding Foresight Ventures and building Foresight News, and then being responsible at Bitget for listing, spot, wealth management, institutional business, and Layer 2. The post presents building the media outlet, the investment firm, and the exchange business as work that unfolded in succession, without giving full start and end dates for each business line.⁠[13]

In a Y TALK interview released in June of the same year, he again listed listing, spot, wealth management, and institutional business, and confirmed that he had also been responsible for Foresight News. He explained that his path began with Foresight Ventures, moved to Foresight News, and then gradually took over businesses at Bitget; some businesses were still starting from zero when he took them over, and he did not manage them throughout their entire development.⁠[5]

On February 7, 2021, Bai Fengxiao relinked his 2019 Bitcoin article in a Zhihu “Thoughts” post and said he was glad of the understanding he had held over the past two years. The post continued his earlier public investment views but provided no trading records, specific investment sizes, or proof of returns.⁠[14]

On July 18, 2021, an Axie Infinity research piece co-authored by Bai Fengxiao and Li Yuanqi was published under the original title “Play-to-Earn Is the Next Liquidity Mining: Is Axie Infinity Overvalued?” The author bios listed the two as a partner and an analyst at Foresight Ventures, respectively. The article compared “play-to-earn” with liquidity mining and discussed the economic incentives created when players own tradable in-game assets.⁠[15]

Using YGG as an example, the two authors analyzed the model in which guilds lower barriers to player participation by renting out NFTs, organize players, and distribute earnings, and also discussed guilds’ network effects and ability to migrate across games. They argued that this model could offer new opportunities to people with limited sources of income.⁠[15]

The research also noted that if Axie breeding outpaced demand from new players, oversupply of assets and SLP issuance could depress prices; guild migration could also lead to player churn. The authors were optimistic about the long-term model at the time and expected the industry to gradually lower yields and improve playability. The return estimates in the article depended on assumptions such as token prices and in-game output and were research judgments of the time.⁠[15]

On August 2, 2021, Matrixport announced the completion of its Series C round at a valuation of more than $1 billion, with Foresight Ventures among the participating investors. The company’s announcement said it had raised a cumulative $129 million since its founding; that figure is total funding, not the amount invested by Foresight Ventures alone.⁠[16]

A related column article published by Foresight quoted partner Bai Fengxiao as saying that blockchain development would change value networks and bring new asset management needs. He praised Matrixport’s capacity for product innovation and said Foresight Ventures would support its global expansion as a long-term partner. The material records his speaking on behalf of the firm at the time and does not disclose any funds he invested personally.⁠[12]

At an industry event in September 2021, he discussed NFTs, game assets, and content life cycles as Foresight Ventures Partner WhiteForest. A transcript of the talk published in October shows that the online name WhiteForest was already in public use in the firm’s early days. The views in the talk were personal judgments under the market conditions of the time and do not represent how NFTs or blockchain games later actually performed.⁠[17]

In the joint report by BCG, Bitget, and Foresight Ventures, Forest Bai was included among the authors and acknowledgments as a co-founder of Foresight Ventures. The report discussed differences between institutional and individual crypto asset allocation and forecasts of user adoption; these forecasts depended on the models and assumptions of the time.⁠[11]

On December 28, 2022, CSI, MicrosoftMicrosoftMicrosoft is a technology company founded by Bill Gates and Paul Allen in 1975 and headquartered in Redmond, Washington, with businesses spanning Windows, Microsoft 365, Azure, enterprise software, artificial intelligence, gaming and devices.Open the full entry’s strategic incubator for Greater China, held a Web3-themed event. A recap credited to “Microsoft Technology” and published by Foresight News in January 2023 introduced Bai Fengxiao as the founder of Foresight Ventures and ForesightX, and recorded that he discussed the industry environment, the problems founders faced, and the firm’s action plans from an investor’s perspective.⁠[18]

The event recap said Foresight Ventures had already established the ForesightX Web3 incubator, aiming to build a global developer network and provide systematic incubation support; jointly building an open-source community with CSI, co-incubating projects, and holding workshops and hackathons were listed as ideas for next-step cooperation. The record describes the incubation direction at the time and does not list completed projects or results from these proposed collaborations.⁠[18]

In a long interview with Web3Brand, Forest Bai explained that Foresight first built its VC arm and then expanded into media and an accelerator; he regarded media as infrastructure connecting Eastern and Western markets and helping overseas projects understand Asia-Pacific users. On investment approach, he said he did not chase short-term hype, placing more weight on leading projects in vertical sectors and on teams’ understanding and focus, and paid attention to AI and blockchain, gaming, and innovative consumer scenarios.⁠[7]

The Information reported that Foresight Ventures acquired a majority stake of about 80% in The Block, paying about $60 million, corresponding to a valuation of about $70 million; Larry Cermak would remain CEO, and Forest Bai was set to become executive chairman of the board. The deal came after the former controlling shareholder of The Block’s loan relationship with Sam Bankman-FriedSam Bankman-FriedSam Bankman-Fried (SBF) is an American entrepreneur and the co-founder of the cryptocurrency exchange FTX and the crypto hedge fund Alameda Research. After FTX collapsed in November 2022, he was charged in the United States with misappropriating customer funds; in November 2023 a jury found him guilty on seven counts of fraud and conspiracy, in March 2024 he was sentenced to 25 years in prison and ordered to forfeit about $11 billion, and in 2026 the U.S. Court of Appeals for the Second Circuit rejected his appeal.Open the full entry came to light and that shareholder stepped down. The equity acquired, the company’s valuation, and Bai’s personal wealth are distinct concepts.⁠[19]

In a December 2024 interview with Metaverse Post, Forest Bai named the combination of AI and crypto, the tokenization of real-world assets, and the bridging of DeFi and traditional finance as areas of focus for the next phase, and positioned Foresight Ventures as a bridge connecting projects and markets in the United States and Asia. The article is an edited interview representing his investment judgments at the time; it is not a complete disclosure of project selection criteria or a promise of returns.⁠[20]

After interviewing more than ten current and former employees, Blockworks reported that several anonymous sources called Forest Bai Morph’s “shadow CEO,” saying he was involved in budgets, strategy, and meetings; the report also recorded that then-CEO Cecilia Hsueh described him as an advisor providing advice on behalf of investors Foresight Ventures and Bitget. Bai did not respond to the report’s request for comment. These claims are not formal findings by a court, regulator, or the company.⁠[21]

In an interview released on June 9, 2026, the host asked whether he had left Bitget “last November.” Forest Bai corrected this, saying he left around June and probably completed the handover and transition by September. Given the year context of the exchange, this is a recollection of his 2025 departure; June and September are approximate times he gave himself, and no specific dates were disclosed.⁠[5]

He explained his departure as a result of changes in the company’s stage and his own: as the exchange grew, it needed more seasoned professional managers, while he wanted to start his own venture in his early thirties. Although other companies later invited him to join, he ultimately chose to start his own business, and he regarded bearing risk, making judgments, and being responsible for a team as a different experience from serving as an executive at a large platform.⁠[5]

In August 2025, Foresight Ventures announced that Alice Li and Zac Tsui had been promoted to partner, and said co-founder Forest Bai would step down from his management role and continue to support the firm as an Advisor. The current official website team page still lists him as Co-founder & Advisor, so “becoming an advisor” does not mean his historical co-founder status was revoked, nor does it mean he completely cut ties with all portfolio projects or media assets.⁠[22]

On October 3, 2025, answering a Zhihu question on the topic of hi-fi, Bai Fengxiao recalled his experience using a variety of headphones and playback devices. He said that because he frequently traveled by air for business, he later came to value wearing comfort and active noise cancellation more; for home audio, he preferred to set up simple systems that connect easily to streaming and home theater, and valued the look of equipment and convenience in daily use. These are personal hobbies and usage preferences he has shared publicly.⁠[23]

On March 18, 2026, Forest Bai posted that this was the first time, about half a year after leaving Foresight Ventures and Bitget, that he was publicly explaining what he was working on. He summarized his new direction as agentic trading products that help traders improve their capabilities, and rejected understanding the field merely as a strategy-selling platform or as autonomous trading programs that fully replace people.⁠[24]

A long post published the same day further proposed that investors retain intent, judgment, and risk preferences while agents provide support for execution, memory, and discipline. Using the example of reducing a position by half across exchanges, he illustrated how data collation, order splitting, and hesitation in the moment can hinder executing a plan; he also listed observability, predictability, and controllability as product requirements, emphasizing the tension between probabilistic large models and low-latency, high-reliability trading systems. These are his entrepreneurial claims, not verification of product performance.⁠[13]

On June 9, 2026, Y TALK released an interview of about 34 minutes with Forest Bai. The program’s opening introduced him as the founder of MOJO AI, showing that the project existed before the product publicity in August. In the interview, he said the company had been operating for more than five months at that point; the program did not specify the recording date.⁠[5]

On team and funding, he said that at the time there were no co-founders and no funding had been raised yet, and that he was mainly using his own money. He wanted to reduce the cost of repeated internal alignment, while encouraging the team to experiment in areas he was unfamiliar with, and to strike a balance between centralized decision-making and delegation. He acknowledged that funding helps with experimentation, hiring, and establishing a value reference for employee options, and planned to consider fundraising after the product had been tested by the market, rather than permanently rejecting outside capital.⁠[5]

He described MOJO as an agentic trading operating system for individual investors: users remain responsible for judgment, while the system assists with research, data analysis, strategy generation, and trade execution. His explanation of “generalization” covered three levels: workflow steps, asset classes, and trading intent, with examples such as announcement monitoring and funding-rate arbitrage; these examples are product design descriptions and do not mean all strategies and markets have been officially opened.⁠[5]

On “personalization,” he hoped the system would, through ongoing interaction, understand users’ preferences regarding asset types, holding periods, leverage, risk, and market capitalization, forming an updatable memory. He also proposed moving from passive question-and-answer to proactively discovering opportunities that fit users’ preferences, storing key memories in a database and looking back at the original context when necessary.⁠[5]

On the technical approach, he argued that trading systems must prioritize execution accuracy, timeliness, and a very low margin for error, and cannot simply add a chat window to an existing platform. Using the example of wrapping exchange APIs and their parameters, he explained that the team needs to prepare callable execution tools for large models; improvements in model capability do not automatically take care of this interface and reliability work. His statement about “100 percent accuracy” is a requirement for the execution goal; the interview provided no test results showing that the goal had been achieved.⁠[5]

Asked whether it could help retail investors make money, he explicitly denied that any product can automatically guarantee profits and did not promise that using MOJO would be profitable. He explained the goal as providing ordinary traders with better research and execution tools, reducing cases in which they deviate from their original plans amid cross-platform order placement, fatigue, and hesitation; users can still change instructions or stop execution.⁠[5]

On trust mechanisms, he set out three requirements: users can understand in advance what the agent will do next, can view the actions already executed and the reasons for them, and can modify or shut down tasks at any time. In response to the host’s concerns about employees spying on strategies or planting backdoors, he mentioned using models to assist with code audits and considering adopting learning mechanisms to protect privacy in the future; these are methodological claims and plans from the interview and cannot be taken as proof that internal abuse has been eliminated or that an independent security audit has been completed.⁠[5]

He said the product aims to focus on the value of a user’s entire life cycle rather than merely encouraging more trades, and envisioned identifying and flagging impulsive trades. On competing with large exchanges, he argued that small teams are more flexible in organizational and product transitions, and that user preferences and memory accumulated over time can reduce willingness to switch; he also cited cross-platform service, not limited to assets already listed on any one exchange, as the differentiation he envisioned. These are his business judgments.⁠[5]

Looking back on his investment work, he denied taking a spray-and-pray approach, saying the investments at the time numbered roughly a few dozen projects, and mentioned the influence of Duan YongpingDuan YongpingDuan Yongping is a Chinese entrepreneur and investor, born in 1961 in Nanchang, Jiangxi, who graduated from Zhejiang University in 1983. He built the Subor and BBK brands, co-founded Guangdong BBK Electronics in 1995, turned to investing in 2001 and in 2006 won a charity lunch with Warren Buffett. He has made large donations to Zhejiang University, and from 2026 he publicly said he was buying shares in Pop Mart.Open the full entry, Feng Liu, and Sequoia on his approach. He summarized his selection method as finding the strongest team in sectors he was bullish on; the roughly $400 million discussed in the program was the firm’s combined assets under management across primary and secondary market businesses and cannot be regarded as his personal assets or investment returns.⁠[5]

He also placed great importance on focus and execution in entrepreneurship: abundant resources do not necessarily lead to better decisions, and a small team concentrating on completing one product may be more effective; faced with multiple viable options, he tends to first filter out priorities by weighing input against output. He regards MOJO as an undertaking that will need to last more than ten years and, beyond commercial returns, hopes to take part personally in the process of AI changing how people interact with trading.⁠[5]

On July 28, 2026, Bai Fengxiao posted a hiring notice on Zhihu for a Trading Agent team in Hangzhou, saying he was building an AI trading startup team in Hangzhou and hoped to give ordinary investors tools similar to those of quantitative teams. The post did not name MOJO directly.⁠[25]

The hiring post described the product’s tasks as understanding market conditions, news, on-chain data, and users’ holdings, proposing trading plans and risk alerts, and executing after user confirmation. Drawing an analogy to Claude Code’s assisted programming, he stressed that the system should be able to understand context, break down tasks, call tools, and verify results, rather than merely providing fixed workflows or ready-made strategy lists. These were product ideas at the time of recruitment.⁠[25]

The positions included front-end, UI/UX, back-end, and testing, with a published annual compensation range of $100,000 to $300,000 plus options, depending on ability and fit. He also said team members came from Tsinghua University, Peking University, Zhejiang University, the National University of Singapore, and University College London, and that a former CMO of Futu and an AI professor at the National University of Singapore had already joined. All of the above team and compensation information comes from his hiring post; the post did not disclose the names of the people involved, their specific roles, or actual payroll records.⁠[25]

On August 25, 2026, CoinNess reported on Forest Bai’s plan to launch MOJO, describing it as an AI trading app integrating data analysis, research, investment judgment, and trade execution, covering scenarios such as spot, real-world assets (RWA), and perpetual futures. The report said the project planned to begin limited testing from August 27 and recruit participants through a waitlist. The report records the testing arrangements at the time.⁠[26]

On September 10, 2026, Forest Bai appeared as a guest on an X Space hosted by 168X and moderated by Mr. Z, titled “Conversation with Forest Bai | Mojo AI: Turning Trading Ideas Directly into Executable Agents.” The event was announced for 19:00 Hong Kong time and noted as sponsored by Mojo AI; the public page later showed that the event had ended, with the recording card showing a duration of 1 hour 38 minutes 26 seconds.⁠[27][28][29][30][31]

According to an interview write-up later published by the organizer and written by the host, Forest Bai judged on the program that Bitcoin was nearing a cycle bottom, but that as the asset grew larger and AI attracted capital and attention, one could no longer expect each cycle to bring the high-multiple rallies of the past. He still leaned toward holding Bitcoin and expected the bottom to come within three to six months from that time. He was also optimistic about the long-term opportunity in Chinese technology stocks, believing that the relevant entrepreneurs and companies were undervalued. These were his investment judgments at the time of the interview.⁠[32]

On decentralized perpetual futures exchanges, he argued that the simpler the protocol, the stronger the network effects and the harder it is for competitors to differentiate; he attributed HyperliquidHyperliquidHyperliquid is a layer-one blockchain designed from scratch for financial trading. It uses its own HyperBFT consensus, has perpetual futures and spot order books built into the chain (HyperCore), and offers an Ethereum-compatible smart contract environment (HyperEVM); its native token is HYPE. Launched in 2023, it became one of the largest decentralized perpetual futures venues thanks to its fully onchain order books, and later opened permissionless deployment of perpetual markets on commodities, stock indices and FX, as well as outcome markets, through HIP-3 and HIP-4.Open the full entry’s advantages to its trading performance, liquidity, token community, and design for open market deployment. For teams planning to enter the field, he advised prioritizing broader coverage of tradable assets, arguing that gaining market access is more fundamental than early liquidity. He also expressed reservations about the long-term business potential of prediction markets, believing their main opportunities lie in long-tail demand, and valued compliant entry into the U.S. market.⁠[32]

He said that after leaving Bitget and Foresight he had considered continuing with a perpetual futures platform or compliance-based business, but ultimately chose an AI trading startup, hoping to change how people interact with traditional trading interfaces. Looking back on the earlier boom in digital asset treasury companies, he said he had not taken part in related investments and questioned whether such structures created new value; this statement was not accompanied by a complete list of personal or institutional investments.⁠[32]

In describing Mojo AI’s product, he stressed letting users state their trading intent clearly in natural language, with the system generating executable scripts that, after user approval, continuously fetch data, check conditions, and execute. He said the actual execution stage is programmatic and does not rely on step-by-step large-model reasoning, in order to meet trading’s accuracy requirements. His examples included recurring purchases, splitting orders over time, funding-rate arbitrage, and using prediction-market signals to trigger trades on another platform; these are the interviewee’s descriptions of product features and implementation and have not undergone independent performance or security verification.⁠[32]

He positioned Mojo as a top-level entry point connecting multiple trading venues, said it does not custody user funds, and planned to prioritize integration with centralized exchanges so that users can manage their accounts while their funds remain on the original exchanges. Priorities for the next three to six months mentioned in the interview also included improving user onboarding, adding convenient menus, and optimizing the interface; further compliance-based business and letting users trade real stock assets were later directions. These integration and compliance arrangements were plans as presented on the program.⁠[32]

On market expansion, he said social media and content promotion would use global English-language communication, while offline business development would first focus on China, South Korea, and Southeast Asia, with part-time promotion partnerships open. He treated partner revenue sharing and user growth as early priorities, hoping to expand the user base through simple incentive arrangements; the write-up provided no verifiable data on how many paying users or how much revenue these channels had brought in.⁠[32]

The business model he described included AI subscriptions and some trading fees: subscriptions have a free tier and tiers at $29, $99, and $399, with no charges during the beta; no additional fees are charged for trades through centralized exchanges, while fees are charged in some other trading scenarios. The write-up did not specify the billing period for these subscription amounts. He said he hoped to grow over the long term on revenue and profit and to pursue a listing when regulatory conditions are suitable, and had no interest in issuing a token; these were his business arrangements and goals at the time and do not mean the company has listed or has made a permanent commitment never to issue a token.⁠[32]

As of September 2026, Forest Bai’s own X account showed the display name WhiteForest and the handle @ForestBai1, with a bio listing experience including MOJO, Foresight Ventures, The Block, and Bitget and linking to the Foresight Ventures official website. The firm’s official website lists him as co-founder and advisor; the Y TALK interview introduced him directly as founder of MOJO AI, and he said that at the time of the interview there were no co-founders.⁠[33][5]

His Zhihu profile uses the Chinese name Bai Fengxiao, lists Foresight Ventures Co-founder in its bio, and displays a badge for a master’s degree from the Department of Thermal Engineering at Tsinghua University. In September 2026, the profile used a portrait photo with a mountain background. The firm and education information on the account also corresponds with the aforementioned interviews and university record.⁠[34]

In a 2020 answer about online usernames, he said he had always used his real name.⁠[35]