Dramatic Shift in AI Startup Requirements: Must Earn Revenue Before Securing Funding
-
How much smaller is the margin for trial and error for teams entering the AI field now compared to the mobile wave years ago?
At least 80-90% smaller. Previously, venture capital had much higher tolerance, and investment decisions were easier to pass, but the situation has changed significantly now.
So what should AI startups focus on today?
Revenue! You must earn money first."
During a phone discussion with Wu Shichun, founding partner of Plum Ventures, we talked about the current super-hot AI startup and investment scene.
If we were to summarize the key takeaways, two aspects stood out most:
First, the dramatic changes underway. Today's entrepreneurial and investment climate is completely different from the past. The once-tolerant space for trial and error has been rewritten by current market conditions, with permissible error margins reduced by at least 80-90%.
As a result, not only has the approval rate for investments sharply decreased, but the required time has also significantly lengthened. The demand for startups to first validate their business models and revenue potential—or even to earn money first (even if small amounts)—has suddenly become much higher.
In one sentence: Startups must generate revenue from the outset.
On the other hand, there are opportunities hidden behind these dramatic changes. For example, we briefly mentioned the demand for cost reduction and efficiency improvement for merchants, the need for applications supporting automated marketing and operations for various teams, and the new dynamics between local governments, VCs, and startups.
To some extent, providing applications and services for merchants can generate revenue and grow faster. This is also the original intention and reason behind the AIGC Strong Applications Conference jointly initiated by Plum Ventures, Jianshi Technology, and Xingxing AI, which will be held in Beijing on September 20.
The three teams aim to promote the development and application of AI, facilitate face-to-face communication between investors and startups, and showcase new applications to merchants. Wu Shichun and the Plum Ventures team will be present at the conference and the evening secret meeting. Everyone is welcome to sign up for in-depth discussions at the end of the article or through the official account menu.
Now, let’s return to the conversation with Wu Shichun and listen to his thoughts and judgments on entrepreneurship in the AI field:
Jianshi: The enthusiasm for entrepreneurship in the AI field is very high, but we’ve noticed that investments are much more cautious than before. There’s a lot of looking but little action. Is this observation accurate?
Wu Shichun: Although AI entrepreneurship is the top area of interest for institutions, actual investment activity remains limited. Many institutions are willing to explore, engage, and discuss, but they remain relatively cautious when it comes to making actual investments.
Regarding AI applications and enterprise service startups, I believe investments will primarily focus on companies that already have established scenarios, customers, and data, combined with AI applications. Pure AI-native teams raising funds are relatively rare.
Jianshi: If that's the case, AI applications and the enterprise service market are closely intertwined. In recent years, the capital market hasn't shown much enthusiasm for enterprise services and SaaS, especially as the sector has cooled. If AI applications move in this direction, could it change the previous lack of capital interest in SaaS?
Wu Shichun: I think part of it will definitely change, potentially revitalizing SaaS. There was already a major wave of investment in SaaS, but the returns for investors were mediocre. This wave of AI brings opportunities like cost reduction, efficiency improvements, and expanded services for customers.
Back then, SaaS investments were often benchmarked against overseas counterparts, with valuations based on U.S. listing standards, dividing by 10 or 20 to estimate Chinese equivalents.
But I believe AI applications may not necessarily follow the same overseas benchmarks. During the SaaS era, many aimed for U.S. listings, but now AI companies are more likely to list in the A-share market. A-share listings prioritize profitability metrics, making comparisons with U.S. AI applications less relevant.
Therefore, in AI application investments, the emphasis is on becoming the first to list in China, where each step can be benchmarked against existing information, making it easier to proceed. However, when no benchmarks exist, one must independently explore pricing logic, future listing expectations, barriers, and review processes, requiring more due diligence (DD) research.
Jian Shi: What angles and aspects are currently prioritized in research?
Wu Shichun: I believe many projects in the financial sector, vertical applications that can generate profits quickly, and areas like recruitment, legal services, and AI education—where education remains a significant market—are key focuses.
Jian Shi: From your perspective, how has corporate willingness to pay changed compared to earlier SaaS models?
Wu Shichun: Overall, willingness to pay has not improved significantly, which is also related to the industry's current economic conditions. Companies are likely cutting costs, reducing unnecessary expenses, and minimizing non-essential applications. This has a major impact on the current paid market.
For AI, if it can help customers generate more revenue than previous SaaS models and become a tool for increasing income, I believe such tools can attract greater willingness to pay from customers.
Jian Shi: The macro-environment of 'cost reduction and efficiency improvement' for enterprises is unlikely to change significantly in the coming years. Could this delay the progress of AI entrepreneurship by several years?
Wu Shichun: I believe there's currently a lot of noise and hype, but for companies to truly achieve profitable effects, they need to demonstrate scalable revenue, IPO capability, and the ability to deliver returns to investors.
In this process, everyone is waiting for a breakthrough. Once such a breakthrough is achieved—like when Sina, Sohu, and NetEase went public—it will serve as a landmark event that inspires entrepreneurial enthusiasm even more than the earlier wave of replicating U.S. business models in China.
Therefore, we need such success stories now. Indeed, people need to make money from it, and the same applies to AI development.
A reference point in this industry is OpenAI, which has achieved annual revenues of $1 billion, a market valuation of tens of billions, and is rapidly approaching hundreds of millions of users.
These benchmark achievements will significantly attract talent, capital, and societal attention. China also needs such landmark companies and events.
Jian Shi: There is indeed a challenging issue here.
First, large models have become the stage for big platforms and companies. Second, in the enterprise services market, platforms are increasingly connecting directly with countless merchants. The service provider market has been significantly suppressed, with even listed companies becoming rare in the new ecosystem in recent years.
This trend extends to AI, where platforms may encourage merchants to collaborate and pay directly. In this scenario, how much room is left for startups focusing on these large models and enterprise applications?
Wu Shichun: In reality, if we consider AI as a wave and opportunity larger than the internet and mobile platforms, many things have yet to begin.
Thus, as its prologue just unfolds, what we can predict may be far less than what will eventually materialize.
It’s like when you first used email—you might have exhausted your imagination but still couldn’t foresee the vast impact and utility the internet would bring. Investors, therefore, must learn, compare, and gradually place their bets.
This sentiment is a mix of fear of missing out and fear of making wrong bets. Amid this conflict, investors are willing to observe, inquire, learn, and eventually muster the courage to invest.
Jianshi: Earlier in venture capital, funds were relatively abundant, and there was more tolerance for trial and error. But in recent years, the entire VC landscape has become challenging. In such an environment, how much room is left for trial and error?
Wu Shichun: Currently, the primary funding comes from government support. The government focuses more on industrial implementation and tax contributions for new projects, which current AI application companies may not yet be able to provide.
In reality, the room and opportunities for trial and error for these startup teams are much smaller compared to the mobile wave a few years ago. Back then, there was abundant social capital and dollar funding, with higher tolerance and easier investment approvals.
Jianshi: How much has it decreased?
Wu Shichun: By 80% to 90%.
Jianshi: If the initial requirements for revenue generation, profitability, and customer satisfaction are very high for many AI startups—meaning they can start making money right away—would their financing valuations or willingness to invest be completely different?
Wu Shichun: It depends on the scale of the earnings. If they make significant profits, of course, things would change.
Of course, making small profits only proves you have the ability to generate revenue, but the amount may not be enough to support scaling up. So, even with small profits, startups still need financing, capital recognition, and amplification to grow.
We prefer to see entrepreneurs who can generate small profits early on, demonstrating the ability to be self-sustaining. In the AI field, we often say: hone your skills on small opportunities to capitalize on big ones.
You may need to sharpen your skills persistently, like an indomitable cockroach, until a major AI application comes along that allows you to make significant profits. Companies like Tencent also honed their skills through various small ventures, such as early SP revenue, before dominating the gaming market.
However, making money doesn’t mean a company no longer needs funding. I believe no company will achieve exceptionally large profits in the near future.
Jianshi: As the demands for revenue and customer capabilities grow heavier, will new forms of collaboration emerge in financing, entrepreneurship, and industry coordination?
Wu Shichun: Very likely. The investment paradigm is shifting, leading to more suitable relationships between investors and entrepreneurs, as well as between VCs, governments, and LPs. Everything is in flux, and we are merely small participants observing the subtle pulses of change.
Jianshi: Looking at overseas markets, some small teams of just a few people are achieving strong revenues in niche verticals by leveraging automation tools for management, marketing, and services. The future may see a rich diversity of combined applications.
Yet, it remains unclear whether the rise of small teams and entrepreneurship will become a reality, and whether applications tailored for small teams will flourish.
Wu Shichun: I believe this sector will grow, and I'm very optimistic about it. We've reviewed over a dozen projects and are currently conducting research and due diligence.
In fact, the time to complete investments has slowed compared to before. For example, projects identified in May or June might not close until August or September if we decide to invest.
Jianshi: In that case, can many teams survive this extended period?
Wu Shichun: You could view this as part of VC evaluation criteria. If a team can't endure this phase, they likely wouldn't pass the first round of screening anyway.
Jianshi: If startups can generate revenue quickly in the new environment, should they skip straight to Series A funding? Will early-stage rounds like seed or angel funding disappear from the market?
Wu Shichun: Many of our current investments are at Pre-A or Series A stages, with a slight shift toward later rounds. Typical investment amounts now range between 2-3 million RMB.
In the past, there were indeed many seed/angel-stage projects, but current valuations and funding amounts are significantly higher when reviewing overall data.
Jian Shi: With the restructuring of the Plum Blossom team, what new requirements are there for the research team?
Wu Shichun: Our work needs to be more meticulous. The cost of researching a project is much higher now compared to before. In the past, decisions could be made in minutes, and investments could be finalized within days, but that’s no longer feasible.
Jian Shi: How do the current AI entrepreneurs differ from those in previous years in terms of requirements and overall capabilities?
Wu Shichun: I’ve proposed a concept called "MSHTC," which represents a strong team composition: M for Visionary (Dream Builder), S for Executor, H for Industry Veteran, T for Technical Leader, and C for Content Head. A team must excel in sales, technology, content, management, and industry expertise.
Therefore, the requirements are now more comprehensive, with no room for weaknesses.
Jian Shi: Are such ideal teams common now?
Wu Shichun: At the very least, one should possess awareness in this regard and strive to proactively address such challenges.