Since Chen Yiyang agreed, he promptly said his goodbyes and prepared to return to Ludao City to start getting ready.
Although Kuxing had already gone public on the Nasdaq once, the previous departure wasn't very glorious, so this time going back would take quite some effort.
After seeing Jiang City off, Chen Yiyang glanced at his phone.
He found several missed calls from Wen Liangfeng.
So Chen Yiyang quickly called Wen Liangfeng back.
"Has something urgent happened at the company?" Chen Yiyang asked.
"No urgent matter, I just wanted to tell you something. Earlier, I mentioned wanting to take DeepSeek public, but now I've changed my mind and don't plan to proceed for now."
Day after day.
Chen Yiyang recalled that Wen Liangfeng had indeed mentioned plans to list DeepSeek to raise funds and accelerate the expansion.
However, they hadn't discussed details, so Chen Yiyang hadn't paid much attention, just had his assistant do some preparatory work.
Now, Wen Liangfeng was regretting his decision.
Actually, Chen Yiyang wanted DeepSeek to go public because the current competition among AI companies is essentially a money-burning contest.
Taking DeepSeek public could alleviate Chen Yiyang's financial pressure.
Who would have thought, he'd be forced to list when he didn't want to, and when he did want to list, plans changed.
"Did something happen to make you suddenly decide against going public?" Chen Yiyang asked.
"The current climate is unfavorable; a large number of investors are preparing to short AI stocks. I want to wait until the storm passes," Wen Liangfeng replied.
"Shorting AI stocks, what exactly is going on?" Chen Yiyang asked.
"In North America, there's a movie about the 2008 Subprime Crisis called 'The Big Short.' Have you seen it?
In the film, several people had shorted the stock market before the crash and made a fortune.
One of the real-life characters is now an investor, Michael. He's recently targeted the AI sector, accusing several major North American tech companies of using accounting tricks to inflate profits amid the AI frenzy.
Thus, several top Wall Street bank executives and renowned short sellers have issued warnings, expecting a possible correction in US stocks.
Going public now might negatively impact DeepSeek."
Accounting is a very interesting field.
Although numbers don't lie, accountants can manipulate the order of the numbers to create a misleading effect.
In this industry, those who excel at this skill tend to further their studies at Tilanqiao.
Don't misunderstand, Tilanqiao is not a school; it's the name of a prison in Shanghai.
It's associated with accountants because many accused of financial fraud and falsifying accounts have ended up incarcerated there after their deeds were uncovered.
Therefore, Tilanqiao has another nickname, "Shanghai Finance Branch School."
There's even a rumor that entering Tilanqiao is akin to undergoing further studies. Upon release, one's job prospects might improve.
Chen Yiyang was unsure of the veracity of this rumor, as he requires incoming employees to have no criminal record, meaning no accountants in his company have undergone further studies at Tilanqiao.
However, many top domestic securities firms don't require a no criminal record certificate upon recruitment.
These top securities firms have extremely stringent hiring processes, with thorough background checks.
But they don't require submission of a criminal record certificate.
This fact alone reveals certain truths.
Michael's criticism of AI companies is primarily based on their use of accounting techniques to artificially inflate profits, making financial reports look attractive and misleading investors into investing.
The tactics used by these AI companies are simple indeed.
The primary costs for AI companies are personnel, electricity and broadband costs, and hardware investments.
The first two are hard to falsify. But the last item, while seemingly difficult to fake,
Given the transparency of chip and server prices in the commercial market.
Yet all AI companies use the same trick to reduce these expenses on financial statements.
Specifically, although chip and server prices are transparent, these hardware facilities have a lifespan.
Their typical lifespan is three to five years.
But this is where the companies' crafty maneuver comes in.
While the theoretical lifespan is three to five years,
Accurately assessing actual usage duration is difficult for non-experts.
If these hardware devices are used personally,
Some individuals can even go ten years without upgrading equipment.
Thus, AI companies intentionally extend their usage lifespan.
Originally used for three years, these devices would be scrapped, requiring new purchases.
If the device cost is 120,000, being scrapped after three years, the annual depreciation is 40,000.
But with accounting manipulation, the lifespan extends to ten years, reducing annual depreciation to only 12,000.
Calculating this way, each device saves
28,000 on books annually.
Hardware costs are the largest expense for AI companies.
If a device saves 28,000 annually, an AI company with at least millions of devices.
Simply adjusting depreciation can multiply the company's profits several times.
And investors won't question why equipment lasts for ten years, yet hardware investments remain substantial annually.
Because all AI companies are currently expanding, requiring significant yearly investment in equipment.
Thus, investors perceive AI companies as investing heavily now, but profits remain considerable after deducting initial expenses.
Due to this, many investors have channeled funds into the AI sector.
"Do you know the scale of investment?" Chen Yiyang realized the severity of the problem.
If true, the AI industry could face its first Internet winter.
The Internet winter in the Millennium made survival difficult for all Nasdaq internet companies.
And the unfortunate OnePlus listed on the Nasdaq at that time.
On the first day of listing, its stock plummeted to below one US dollar.
Two years later, OnePlus shares reached seventy US dollars, a hundred and ten times the lowest point.
The Internet winter's impact on companies back then was evident.
If OnePlus hadn't sustained itself, subsequent stories might not have unfolded.
"AI companies changing depreciation lifespan is an open secret in the industry.
As far as I know, Oracle inflated by at least twenty-six percent, Meta by twenty percent, with other companies similar."
This is really terrible.
Chen Yiyang's eyelid twitched hearing this figure.
Such a scale of inflated profits, if confirmed, is likely to cause a loss of investment confidence.
All AI companies would suffer serious setbacks.
Although these are competitors.
If a market, all competitors suffer due to poor reputation, the remaining one would also be scrutinized.
"The good news now is the chip companies providing hardware for AI don't want AI firms to collapse.
Microsoft stated, the chip supplied to AI companies indeed doesn't have just a three-year lifespan.
A chip performs best in the first three years, supporting the most advanced operations for AI companies.
After three years, only performance declines, not complete failure, still useful for AI's other operations, like cloud databases, video transcoding, etc.
It's now up to Michael to present evidence challenging Microsoft's statement."
