However, I believe that a large part of the AIGC boom we are seeing is an illusion. If adjustments are not made in time, the industry will inevitably be in a state of wailing in 2 years.
Of course, I should clarify in advance: I'm not denying the value of AIGC. The implementation of AIGC in certain industries and scenarios is definitely valuable.
However, at least in the next 1-2 years, the value of AIGC in the B2B sector is likely overestimated.
01 Overestimated AIGC
Last year, entrepreneurs in the SaaS Executive Group developed industry-specific AIGC products with positive market feedback, and a major investor expressed investment interest.
But this year he told me:
The biggest value of AIGC products is that they make software sell for more, but in reality, because AIGC-generated content is only 90% accurate, and his field requires 100% accuracy, AIGC products don't actually generate real business value.
As for why customers are still willing to pay, this CEO explained: Customers also need to report intelligent achievements to their superiors, and AIGC obviously appeals to those above.
Another product VP from a leading SaaS company also told me: After ChatGPT was released, they immediately started researching AIGC products, but after more than a year, only 1-2 scenarios actually panned out.
His conclusion is: In their field, AIGC is not yet suitable for large-scale application.
So where does the problem lie?
In fact, it's not that AIGC technology is immature, but that AIGC is essentially just a correlation logic.
For example, it knows that 1+1=2, not because it understands math, but because based on historical data, it infers that there's a 99% probability that 2 will appear after 1+1=, so it gives the result of 2.
But our enterprise management is more about causal logic rather than correlation logic. For example, if a customer buys 2 items, the order amount must be 2 items multiplied by their unit price. This absolutely cannot be inferred by probability.
Everyone can review and see, aren't at least 90% of enterprise business scenarios causal logic?
Such as procurement, sales, inventory, production and manufacturing, financial accounting, supply chain management.
Even some scenarios that don't seem to require 100% accuracy are actually not as casual as we imagine, for example:
A secretary writing meeting minutes, even 1% of critical errors are unacceptable;
A designer creating a promotional poster must 100% comply with corporate UI standards;
Customer service answering customer questions, even 1% of misleading information is unacceptable.
A doctor writing a diagnostic report, even 1% of incorrect conclusions can cause major problems.
So, if AIGC is really used to handle most of a company's business, even if there's only a 1% chance of error, it can bring great losses to the company.
In fact, it's been nearly 2 years since ChatGPT was released, but now we're "surprisingly" still struggling with what scenarios it's useful for?!
Doesn't this say something?
Moreover, even in scenarios where AIGC excels, such as text generation and image generation, in most cases, the effects of AIGC fall far short of enterprise expectations.
You might say that AIGC is constantly evolving.
But no matter how it evolves, it will always be correlation logic, and will never achieve 100% accuracy. This is determined by its genes.
This determines that AIGC can only demonstrate real value in a few scenarios.
But obviously, many people are still unwilling to acknowledge the severity of this problem.
02 AIGC Will Inevitably Face Market Ceiling Issues
Even if AIGC eventually finds suitable business scenarios, I dare say its development in China won't reach the level of Europe and America.
The implementation of AIGC in the B2B sector is essentially enterprise software. So, the path AIGC is about to take, SaaS has already paved the way.
The key to SaaS's rise is the implementation of mobile internet in the B2B sector. It can even be said that:
SaaS = Enterprise Software + Mobile Internet.
This is why 2015 was called the first year of SaaS, because the popularization of mobile internet in 2014 was the biggest driving force for the explosion of SaaS.
Compared to AIGC, the implementation of mobile internet in the B2B sector was very smooth, after all, many business scenarios in enterprises can be mobilized.
Even so, the development of SaaS in China is still far from expectations.
There are two very critical reasons for this.
First, the problem with Chinese SaaS is not a product problem, but a market problem.
Some say Chinese SaaS products are not good, Chinese SaaS companies lack capability and awareness. But even products from big companies like Feishu and DingTalk haven't achieved large-scale profitability.
The main problem with Chinese SaaS is still that customers don't recognize the value of software and have limited ability to pay.
I will explain this point later with 3 key numbers.
Second, the target customer groups for Chinese SaaS and AIGC basically overlap.
This means that the market problems that Chinese SaaS hasn't solved, AIGC will have to face one by one.
For example, Kai-Fu Lee said recently: In China, many companies haven't recognized the value of software and are unwilling to pay for it. Plus, many large model companies participate in bidding, driving prices lower and lower, greatly compressing profits, losing money on every deal.
Another executive from an AIGC startup told me: Customers don't have a strong willingness to pay for AIGC, and projects generally have a high degree of customization, resulting in low input-output ratios, long delivery and payment cycles, making it impossible to sustain the R&D team.
Familiar scenarios, familiar taste?
Let me show you three important numbers, and you'll understand the common problems faced by SAAS and AIGC.
The first number, according to data from the National Bureau of Statistics, China's two largest industries in the first half of 2024 were manufacturing and wholesale and retail, accounting for nearly 40% of GDP combined.
However, these two industries are more traditional, with a low degree of online business, so they don't have high recognition of software.
In contrast, the tertiary industry in the United States is more developed, such as high-tech and finance, accounting for 70% of market entities. Their businesses are mainly online, placing great importance on information processing and collaboration efficiency, and of course, have higher recognition of software.
We can also refer to Feishu in this regard.
Feishu's characteristics are good user experience and high collaboration efficiency, but it's relatively expensive, so those who truly recognize Feishu's value are often enterprises in the tertiary industry such as internet and finance.
Because they are all talent-intensive enterprises.
The second number, according to the "Comparative Study of China-US Top 500 Enterprises White Paper" released by CCID Think Tank in 2021, since 2016, the average profit of US manufacturing companies on the list has been about 4.9 times that of Chinese manufacturing companies on the list.
In other words, even comparing the same industry, the profit level of US companies is much higher than that of Chinese companies.
The higher the profit level, of course, the more willing they are to invest in enterprise software, which is a non-rigid demand.
The third number, according to Gartner's data, in 2021, about 42% of global IT spending was invested in IT services and application software, with only 19% invested in hardware.
In comparison, Chinese enterprises invested 19% in IT services and application software, but 31% in hardware.
In other words, compared to the global level, Chinese enterprises prefer to buy hardware rather than software.
The above 3 numbers actually all point to one thing, which is that Chinese enterprises have weak willingness and ability to pay for software.
And this problem, AIGC will inevitably have to face.
So, don't be superstitious about AIGC, it might succeed in the US because there's a completely different market soil there.
But in China, it might be a different story.
03 What to do
If we acknowledge the problems with AIGC, then starting from today, we should adopt a more cautious strategy for AIGC projects and not follow the old path of Chinese SaaS.
In the early stages, many SaaS companies were profitable, or rather, had healthy business models.
However, excessive desire for capital and disorderly internal competition in the industry eventually led the entire industry into losses.
If AIGC pays attention to this problem now, I think it's possible to avoid such a bad situation.
- Don't over-finance
AIGC still has a long way to go in the B2B sector, the future might be beautiful, but there's still a difficult road ahead.
Be more pessimistic, maintain minimal operations, reserve more time for MVP development.
Financing is okay, but don't fabricate data for financing