In 2025, when there's an abundance of money, will AI cheating companies become the darlings of anxious capital?
In June 2025, the myth of "being expelled from Columbia University but securing tens of millions of dollars in financing" swept through the Silicon Valley venture capital circle. Cluely, an "AI cheating" company that had only been established for two months, received a $15 million investment from a16z, a well - known Silicon Valley VC firm, and its valuation quickly soared to $120 million.
Behind the controversy lies a serious disconnect between AI Agent applications and the capital market: Primary market and VC teams are no longer pursuing technological barriers and long - term growth. Instead, founding teams that "know how to market" are being frantically pursued by capital. Is Cluely's sudden popularity a microcosm of capital's desperation or a precursor to the iteration of the entrepreneurial paradigm? With no sign of a killer application on the consumer side, what will be the new paradigm for entrepreneurship in the AI era? And why do major investment institutions no longer count on IPOs?
As the so - called "Year of AI Applications" that was bet on is entering its second half, this article will summarize the investment situation in the AI circle in the first half of 2025, explore the investment logic of top Silicon Valley funds, and discuss the new trends in AI entrepreneurship.
01 The "AI Cheating" Company Cluely
Cluely's latest round of financing caused quite a stir in Silicon Valley because this product clearly teaches users how to cheat in interviews.
Openly written on the company's website: We just want to cheat on everything.
This AI startup is called Cluely, and its founder is Roy Lee, a Korean - American student who was recently expelled from Columbia University for launching an AI cheating software.
But no one expected that such a controversial young man, whether you think he has no moral bottom line or call him a "marketing madman", could really secure financing.
Moreover, he claims that just 16 days after the product was launched, the Annual Recurring Revenue (ARR) reached $1 million, and it reached $5 million in just two months.
However, such a rebellious and controversial young man managed to secure tens of millions of dollars in Series A financing from the most mainstream funds in Silicon Valley. This really puzzles many people and makes many hard - working entrepreneurs feel unfair: Is this really okay?
But then again, what we are more concerned about is the underlying logic behind this event: More than half of 2025 has passed, and AI applications have not exploded as expected. Are the investors who claimed last year that 2025 would be the "Year of AI Applications" starting to get anxious and desperate, and just chasing any project that can make money? Are VC funds still entering the market?
02 Continued Influx of Capital into the AI Track
Capital is still flowing into the AI track, and the amount and scale are even larger than before. The main reason behind this might be quite ironic: there is too much money.
Let me show you a chart based on data from The Information.
The dark red area on the left: The leading investment amount of VC funds in the AI field from the first quarter of 2022 to the second quarter of 2024.
The light pink area on the right: The AI investment amount from mid - 2024 to mid - June 2025.
It can be seen that top funds have been pouring into the market from last year to this year, especially SoftBank and Thrive. Lightspeed and a16z, the funds that invested in the aforementioned AI cheating company, are also making substantial investments.
Simply put, from an investment perspective, the AI track has not cooled down, and funds still have a large amount of cash waiting to be invested.
Jenny Xiao
Partner at Leonis Capital, former researcher at OpenAI:
If you look at the development of AI in 2023, a mainstream view at that time was that ChatGPT might not be popular for long, and the products launched by MetaAI, xAI, and Claude would surpass or match ChatGPT. However, unexpectedly, ChatGPT has experienced continuous and rapid growth. After noticing this trend, institutions like SoftBank and Thrive doubled down on OpenAI.
The current Silicon Valley venture capital market is in a stage of crazy influx of hot money, especially when the secondary market is full of uncertainties, and the primary market is catching the big wave of AI.
Jenny Xiao
Partner at Leonis Capital, former researcher at OpenAI:
The difference in the AI era is first reflected in the significant increase in the scale of VC investment funds. Five or ten years ago, there were very few VC funds with a scale of one billion dollars. But now, I've seen funds with a scale of one billion or even ten billion dollars. Such funds can only invest in large - scale projects that are all - or - nothing. They won't invest in projects that can only be sold for $200 million or $300 million, as such projects are no longer attractive to many large funds.
This is why many VC funds require companies to raise $5 million, $10 million (five million to ten million dollars), or even more in the seed round. Because for them, an acquisition price of $200 million or $300 million is not attractive at all. They are aiming for a result of $20 billion. This is also why many funds make large - scale investments in star teams in the seed round.
03 Potential Compression of Return on Investment
When there is too much money in the market, the return on investment of VC funds is often compressed. Our guest Jenny Xiao compares the current time to 2020 and 2021.
At that time, the investment hotspots were SaaS (software as a service) companies. The industry atmosphere was that all SaaS companies were raising huge amounts of capital, frantically hiring people, expanding their businesses, and continuously launching new products and features. A large amount of VC capital was poured into the SaaS field, resulting in poor return performance and even losses for many funds in 2021.
Jenny Xiao
Partner at Leonis Capital, former researcher at OpenAI:
I think the core problem is that VC funds have raised too much money. People see some trends in AI currently, and many blame AI companies or the technology. But in my view, the essence is that VC funds have raised too much money, and they have to invest it. So they look for companies that are very capital - intensive but might surpass OpenAI and Anthropic. As a result, nine times out of ten, such companies end up losing money, and the return on investment of the entire VC industry is compressed.
I think the situation from 2024 to 2025 will repeat that of 2020 to 2021. This is a tragedy for VC funds, but it's a great environment for entrepreneurs because it's easy for them to raise a lot of money, especially for AI companies.
As AI development reaches 2025, we are seeing a very disjointed scenario: at the beginning of the year, VC funds were generally optimistic about the explosion of AI applications in this year. However, the reality is that no killer AI applications have emerged yet. Many star AI application companies that were previously favored by the outside world have either gone bankrupt or been acquired by large technology companies at extremely low prices. One of the reasons is that AI startups developing applications are easily swallowed up by large - model companies.
Jenny Xiao
Partner at Leonis Capital, former researcher at OpenAI:
There are many companies developing AI applications, such as AISDR and AIcoding, which might be swallowed up by OpenAI and Anthropic. This is why our fund is very cautious when investing in early - stage projects and pays great attention to avoiding direct competition with large - model companies. However, many large funds, even those with a scale of $1 billion or $2 billion, can hardly avoid direct competition with underlying models. The companies they invest in are directly competing with underlying models, which is why the situation mentioned earlier occurs: the acquisition value might be lower than the total financing amount, and VC funds might lose 50% or even 80% of their investment.
If the underlying technology cannot create differentiation and the startup teams are facing the brutal competition where large - model companies are launching updates almost every week, then what is the moat for startup teams?
At this time, advantages of founding teams such as "good at marketing", "strong execution ability", and "fast speed" have become the qualities that VCs value. Launching products quickly, conducting aggressive marketing, and creating hot topics. Even if the company goes bankrupt, it will be a quick failure. Even if the company is acquired at a low price or through so - called "acquihire", it allows funds to exit quickly. This is the new normal of investment in the AI era, which has fundamentally changed compared to the previous Internet era.
Jenny Xiao
Partner at Leonis Capital, former researcher at OpenAI:
VCs definitely don't want to see acquihire, but most VCs' current thinking is that having an exit channel is better than having none. Instead of letting the company have no exit channel for a long time, it's better to cash out through acquihire as early as possible. In this way, VCs might get their principal back or even get two or three times the return, which is a good return for many VCs.
In the AI era, information and underlying technology are developing so fast that we can tell whether a company will succeed in the first two or three years. This is completely different from the previous Internet and mobile eras, where it might take multiple strategic pivots and a long time to find the product - market fit. But in the AI era, the pace is much faster. We can judge whether a company has venture - backable value and can become a unicorn or a larger company in the first two or three years. So for VCs, a quick exit is a good result. For example, companies like Character.AI produce results in two or three years, which can be considered a "blessing" for the VCs that invested in them.
"Fast" has become the keyword for entrepreneurship in the AI era, not only for VC funds but also for startup teams. People are no longer aiming to become the next Google, Facebook, or Apple. Instead, they are trying to build a Series A or Series B company worth hundreds of millions of dollars in two or three years and get acquired by large technology companies like NVIDIA. Both the teams and the investors can then enjoy a happy vacation. Isn't that great?
To be honest, at least several of the guests we interviewed on our Silicon Valley 101 program in the past year have sold their companies and achieved financial freedom.
Recently, there has been a lot of buzz about Jiahui Yu being poached by Meta for $100 million. He was our guest on the podcast three years ago. (By the way, many of our past guests on the program and at our events have been successfully acquired and exited. So welcome everyone to join Silicon Valley 101 and catch some good luck.)
Looking at it this way, does it make sense that a16z invested $15 million in the AI cheating company Cluely?