news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

the much-anticipated openai financing has finally settled.

the final financing results were amazing:

the financing amount is us$6.6 billion, led by american venture capital firm thrive capital. in addition to microsoft's continued participation, there are also new friends investing for the first time, such as nvidia and softbank group;

the post-financing valuation reached us$157 billion, almost twice the us$86 billion valuation at the beginning of this year, and six times that of musk's xai (xai was valued at us$24 billion after series b financing in may this year).

the various games behind financing are also quite unusual:

according to the financial times, openai hopes to "lock" investors in this round of financing. that is to say, openai hopes investors will not invest funds in its main unlisted competitors, such as xai and anthropic, founded by openai leavers. , safe superintelligence (yes, openai even made a list).

according to openai’s official website, this round of financing is provided in the form of convertible bonds. this also means that the money openai receives comes with its own conditions, that is, the company must complete its restructuring within two years and become a for-profit enterprise. if openai fails to break away from the control of the non-profit board of directors and fails to remove the upper limit on investor returns, the $6.6 billion investment will not be converted into equity, but will become openai's debt - openai will have to pay back the money.

since its inception, openai’s total financing has reached us$18.7 billion. among them, openai has received three financings from microsoft in the past five years, totaling us$12 billion. although new investors want to join, they also need a reasonable withdrawal mechanism. in this case, "reorganization" has become the trump card on the gambling table.

the completion of this key round of financing was announced just after openai held its developer conference in san francisco, usa.

on november 7 last year, openai held its first developer conference. a high-spirited ultraman took to the stage and announced the launch of a gpts store. its largest "funder", microsoft ceo satya nadella, was also invited to the stage to talk about the rise of ai.

just ten days later, openai suddenly experienced a "palace change", and altman was suddenly kicked out of the board of directors and removed from his position as ceo. it was the openai non-profit board of directors that initiated the "palace change".

the "palace change" only lasted a few days, and ultraman made a grand return. butthat farce also exposed the contradiction between openai’s special organizational structure and the commercialization process.

today, it’s the openai developer conference again. but what followed was $6.6 billion in exchange for openai’s determination to restructure.

change does not happen overnight. among these changes, is ultraman's hard work one year after his return. whether it is the developer conference that has become "boring" or the recent wave of resignations of openai executives that has been discussed a lot, they all converge at one point.

the characteristics of "convertible bonds" make openai's financing this time more like a big gamble.

the full name of convertible bonds is convertible corporate bonds, which gives creditors the right to convert the debt into company shares under certain conditions. for investors, this is undoubtedly a guarantee: if it succeeds, the investment will be converted into shares of openai; if it fails, then it will be treated as a loan to openai and they will have the right to withdraw the funds.

this condition is quite critical. it fully illustrates the "stuck point" that investors are most concerned about when it comes to the future of openai. in other words, openai needs to make certain commitments to the future in exchange for real money now.

openai's commitment is to complete the restructuring within two years and remove the upper limit on investors' returns.reorganize, that is, move away from the control of a non-profit board of directors and become a true for-profit enterprise.

at this point, openai's restructuring plan was officially put on the table, and a deadline was set.

openai’s current mechanism is quite special. since 2019, three years after its establishment, openai has formed a limited-profit company under the promotion of altman, who has become ceo, but control remains with the board of directors of openai's non-profit part.

at that time, openai had found that the form of a purely non-profit organization could not support its continued progress (lack of money). especially in 2018, musk withdrew from the openai board of directors and cut off financial donations, which made altman realize that there was a big problem with the non-profit organization model.

however, the mission established when openai was founded is difficult to abandon: in order to prevent artificial intelligence from threatening mankind, actively promote the development of general artificial intelligence (agi) that benefits all mankind, and fight against the monopoly of large companies.

sam altman

by forming a limited for-profit entity under a non-profit organization, openai hopes to bridge the gap between reality and dreams.

setting a return cap for investors is also a product of the "need and need" situation - it is necessary to accept external investment, but also needs to use a return cap to ensure that openai will not rely on external investors. investors can exit through profit dividends, and profits exceeding the return limit are returned to the non-profit part of openai.

after that reform, altman introduced microsoft investment, which really put openai on the fast track of development. chatgpt was only able to come out with the support of technology giants.

the outside world once thought this was a clever design.in october last year, it was reported that openai was negotiating to sell existing employee shares at a valuation of us$86 billion. at that time, the financial times analyzed that such a structure would make openai unable to become a hugely profitable company, and microsoft would have to foot the bill. ; if the $150 billion in profits can be cleared out to microsoft and all external investors, it will completely get rid of microsoft and take back control of the company.

however, the farce a month later changed all optimistic thoughts.ultraman was thrown into a trap by the power structure he designed himself.

in november last year, openai staged a series of farces: first, altman was suddenly kicked out of openai's non-profit board of directors, and his ceo position was also removed. greg brockman, one of the co-founders, ) and left with determination.

within three days, openai was in a turmoil, and almost all employees put pressure on the board of directors, demanding that altman be brought back, otherwise they would all resign. microsoft showed up at the right time, saying it would accept all employees who left openai, and even officially announced that altman and brockman would join microsoft to establish an ai department.

under internal and external pressure, the non-profit board of directors was defeated. the incident ended with altman and brockman returning to openai and the board members being shuffled.

although there are different opinions on the reasons for this "palace change", one thing is certain: openai's non-profit part and the limited for-profit part have serious friction.

this is undoubtedly a blow to altman: the power structure he designed for openai has reached a fork in the road of "choosing one or the other" in his continuous pursuit of the company's commercial development. i'm afraid it's okay to "want both" it doesn’t make sense.

it's also a wake-up call for outside investors interested in openai.

microsoft, which had invested tens of billions of dollars in openai at that time, had neither the right to know in advance nor the right to directly intervene afterward when ultraman was kicked out. it could only cooperate with ultraman on the periphery.

what would ultraman choose?

the evaluation of his berejiklian mentor and y combinator co-founder paul graham seems to have revealed the mystery: “some people stop after making enough money, but sam didn’t seem to be interested in money. very big. one possibility is that he did openai because he liked power more.”

the year of returning to openai is the year that ultraman continues to tighten his power.

the board has been shuffled. only one member of the board of directors who participated in the "palace change" remained. altman became a member of the board of directors again, and the rest were newly appointed members. among them, business talents have increased significantly, including the former executive vice president and general counsel of sony corporation, the former ceo of the bill and melinda gates foundation, etc.

on the other hand, it’s no secret that openai’s executive departures are on a wave.

during this year, among the co-founders, ilya sutskever, who was considered to be the driving force behind the "palace change", left. there are now only three of openai’s 13 co-founders, including altman, brockman and wojciech zaremba. there are only two people who are really active, because for some reason brockman took the initiative to "lie down" and announced that he would take a break, and then fell into a long silence.

among senior executives, 13 have left by 2024. the one who has caused a stir recently is chief technology officer mira murati. she was once a close comrade of ultraman. during the "palace change", she was temporarily appointed as ceo, but internally promoted ultraman's return. .

on the surface, the wave of executive resignations seems to be a thorny problem faced by ultraman, but when you look closely, it is actually more like ultraman's active choice.

the departure of the "old man" is more or less related to the great changes openai is undergoing. and this great change itself is ultraman's active choice.

when mulati left, the information quoted people familiar with the matter as saying that internal divisions within openai had intensified. for example, the dozens of core researchers managed by mulati sometimes carried out plans that conflicted with brockman's plans.

when faced with a turf war like this, or when executives asked for more hiring or resources, altman tended to sit back and watch what happened—allowing the fight to intensify and avoiding making decisions.

in addition, behind the “wave of executive resignations”, what is easily overlooked is the “wave of executive recruitment.”openai has not only accelerated the launch of new products for consumers and enterprises this year, but also introduced new executives with business experience.

for example, chris lehane was hired by openai as vice president of global policy. he is a senior politician who has held a similar position at airbnb. he also served as a lawyer and spokesperson in the clinton white house. he is known as the "master of public relations" . another example is kevin weil, a long-time product lead at twitter and facebook who was hired to oversee openai consumer and enterprise products. and sarah friar, who became openai’s first chief financial officer (cfo) in june this year, was the ceo of nextdoor and helped lead the financing process.

“i’m extremely excited to empower a new generation of leaders.”at the italian technology week event at the end of september, altman stated this in response to the high-level changes.

the so-called wave of executive resignations is more like a major shake-up of openai's top management led by altman: first, to bid farewell to executives who have disagreements about the future of openai; second, to further hold power in their own hands.

after mulati left, altman mentioned in a memo to employees that he would spend more time involved in the technology and product parts of the company, whereas previously he had mainly focused on non-technical issues such as financing, government relations and business cooperate. as a result, mulati did not set up a new chief technology officer, and relevant personnel reported directly to altman.

the reorganization of openai that altman wants to carry out is bound to encounter resistance, among which the "old" executives who are well-known externally and have charisma internally are the most fatal.

as they leave and more fresh blood (business experts) join, the probability of reorganization encountering fatal resistance gradually decreases.

after a year of hard work, ultraman has gradually paved the way for the reorganization of openai, but there are still many challenges ahead.

openai has yet to prove it has sufficient hematopoietic capabilities.

the new york times once reported that openai expects revenue this year to reach $3.7 billion and a loss of $5 billion. this data was also confirmed by cnbc. according to people familiar with the matter, openai’s revenue last month was us$300 million, an increase of 1,700% from the beginning of last year, and sales are expected to reach us$11.6 billion next year.

since the launch of chatgpt at the end of 2022, openai has sold subscription services to consumers, promoted products to enterprise users, and licensed its gpt series of large language models. however, the development cost of openai is high, and it has to increase the purchase of nvidia chips to train and run models.

the outside world has always been very interested in openai's financial status, and investors are naturally worried about openai's hematopoietic ability. in this round of financing, openai needs large investors and has set requirements for the starting investment amount. accordingly, investors also need to know more about openai. allegiscyber ​​capital said in an interview with the wall street journal: “no one would commit $250 million without prohibiting investigations, so i think there must be due diligence. "

for this reason, people have linked apple’s sudden withdrawal from this round of financing to it, wondering whether apple chose not to participate after seeing openai’s financial situation.

secondly, openai tried to reach an exclusive agreement with investors during the financing to prevent investors from investing in competitors. whether this move is really wise is questionable.

on the one hand, although openai has made efforts and even made a list directly, it may not have substantial results.

the information quoted people familiar with the matter as saying that investors are not bound by the contract and there will be no substantial penalties if they invest in competitors in the future. in this case, it is foreseeable that not all openai investors will follow its instructions. one investor who wrote a huge check in the financing revealed that openai's opposition alone was not enough to prevent them from investing in a competitor.

in other words, the "exclusive" consensus between openai and investment institutions will only have a certain effect when openai has an advantage - no one wants to offend the most popular players in the most popular tracks. openai can also use this to expand its advantages and snowball. once openai falls behind or reveals its fatal weakness, the so-called consensus between the two parties can be abandoned at any time.

on the other hand, openai's move not only exposed its own limited confidence to some extent, but also made enemies in the industry.

as mentioned before, it is not uncommon to want to prevent investors from investing in competitors. for example, uber once did this when it prevented investors from investing in competitor lyft, which also did the same thing. but uber investor and former benchmark partner bill gurley believes that these measures may backfire, making investors feel that the company is vulnerable, and instead help competitors attract investment.

companies being included on the list will obviously not be happy. after hearing the news, musk, the boss of xai, who was included in the list, enthusiastically opened the microphone on x (formerly twitter), saying: "openai is a devil."

musk has already taken openai and ultraman to court before, and many competitors such as anthropic and safe superintelligence are inextricably related to openai. during the critical period of transformation and reorganization, ultraman is the least likely to all it needs is more slapstick.

after a year of planning, openai's reorganization has officially begun, but whether it will be a comedy or a tragedy depends on ultraman's follow-up methods.