news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

it’s getting so bad that even intel is going to sell itself?

over the past weekend, there was a rumor that the us chip giant qualcomm might acquire intel. the news was first reported by the wall street journal.

it’s no secret that intel is sick, but it’s still hard to imagine intel being acquired.

intel's stock price has fallen 50% since the beginning of this year. patrick gelsinger has carried out radical reforms since taking over as intel ceo in february 2021, but his idm 2.0 plan does not seem to be going smoothly. the independent chip manufacturing department "intel foundry" lost $7 billion last year, and the meteor laker chip, which is said to represent intel's entry into the "ai era", has repeatedly missed delivery.

gelsinger said the foundry business had encountered difficulties. gelsinger also said that nvidia's success in the field of ai was purely accidental.

all in all, intel's big plan to turn the situation around did not go smoothly, and the ai ​​wave caught it off guard.

the last typical example of a company that was caught off guard by the times was nokia. this brand, which had dominated the mobile phone market for 14 years, failed to anticipate the surge of smartphones. when the first iphone was released in 2007, nokia failed to hear the death knell for it.

interestingly, if we go back 14 years, in 2010, nokia, which wanted to change, briefly cooperated with intel to release the mobile operating system meego. at that time, intel also missed the best opportunity to enter the mobile internet - a few years ago, it refused to manufacture the first generation of iphone.

everyone knows what happened next: nokia fell.

"we did nothing wrong, but somehow, we lost." in 2014, when microsoft announced the completion of its acquisition of nokia, nokia's then-ceo jorma ollila said this, and he cried with dozens of executives present.

intel continues to move forward. missing out on mobile internet is a pity for intel, but it did not hurt it. at the beginning of 2020, intel's market value reached its peak, exceeding $300 billion, which is the combined market value of nvidia and amd.

i wonder if intel, which once wanted to join forces with nokia to compete for an entry ticket to the mobile internet, also believes that nokia "did nothing wrong"?

currently, intel's market value is only us$96.4 billion, while amd has exceeded us$250 billion, and nvidia has reached us$2.85 trillion.

the acquisition is still a rumor. after all, intel has not been sold yet. this 56-year-old company still has room to "turn around". stifel analyst ruben roy believes that the acquisition is "unlikely to be approved by regulators", and axios news network also believes that qualcomm will face huge challenges in acquiring intel. in addition, after the acquisition news, it was reported that apollo global's management was considering investing up to $5 billion in intel.

how to avoid becoming the nokia of the ai ​​era will be a difficult problem facing kissinger and intel.

now that the news that intel may be acquired is out there, the outside world is sighing, but it is hard to say whether it is shocked.

after all, not long ago in august, intel had already "shocked everyone" with its financial report.

how frightening is it? when intel released its august financial report, it announced 15,000 layoffs. as soon as the financial report was released, the market value evaporated by $32 billion in a single day. many shareholders filed a class action lawsuit, claiming that intel "fraudulently concealed the company's problems" and also named ceo kissinger as a defendant.

intel's second quarter financial report for fiscal year 2024 shows that quarterly revenue was $12.8 billion, down 1% year-on-year; the net loss attributable to the company was $1.6 billion, while the net profit attributable to the company in the same period last year was $1.5 billion, turning from profit to loss year-on-year. this loss far exceeded market expectations. the market expected intel to lose money, but the expected figure given was $540 million.

the foundry business, which was highly anticipated, seems to have become an embarrassing burden for intel at this stage.shareholders pointed out that intel's financial report showed that its foundry business was not progressing smoothly, and operating costs were still rising despite declining revenue, which caught the company's shareholders off guard.

since taking office as intel ceo in 2021, kissinger has proposed the revival plan of idm 2.0. the idm model refers to the company's simultaneous design, packaging, manufacturing, and verification of chips. in the past, nvidia designed and manufactured chips through idm itself, which controlled costs very well. idm 2.0 is to consolidate its own chip manufacturing capabilities, while strengthening cooperation with third-party foundries, expanding production capacity, and undertaking wafer foundry and packaging business.

in march this year, intel also spent a huge amount of us$300 million to purchase the world's first new-generation na euv lithography machine from asml, demonstrating its determination to compete with tsmc for advanced process chips below 2nm.

however, in april, a document submitted to the u.s. securities and exchange commission (sec) revealed the price of intel's ambition. as intel continued to increase its foundry infrastructure, the independent chip manufacturing division "intel foundry" suffered losses year after year: $5.2 billion in 2022, $7 billion in 2023, and a loss of $2.8 billion in the current quarter.

the life-saving measures began to "harm lives for money", and people cast their questioning eyes on intel and the key figure kissinger.

“replace the ceo” seems to be a natural call.

after the release of the poor financial report in august, ed snyder, an analyst at market research firm charter, called directly: "intel should look for a reform-oriented ceo from outside and abandon the current ceo kissinger, otherwise the company will be in trouble."

however, was kissinger really wrong?

we still need to go back to the beginning of 2021 when kissinger took office.

at that time, intel's market value was still over $200 billion. in comparison, nvidia's market value was at $300 billion, and amd's was at $100 billion. compared to its peak in 2020, when intel's market value was the sum of nvidia and amd, it has fallen behind, but it is still far from being "far ahead."

intel's problem is also obvious. as chips shrink, semiconductor production becomes increasingly difficult. in terms of advanced chip technology, intel has fallen far behind tsmc, and its 10nm process has been delayed in mass production, and it may even outsource more cpu chips to tsmc.

in the cpu market, amd, as the biggest competitor in intel's x86 architecture system, has acquired more advanced production technology than intel with the support of tsmc and has begun producing 7nm and even 5nm cpu chips, directly threatening intel's core business.

another bad thing is that intel lost apple, a major customer it had worked with for 15 years. a year before kissinger took office, apple stopped using intel processors on mac computers and switched to its own chips based on the arm architecture.

all of these were traps dug by several ceos before kissinger.

after intel's founder and before kissinger, intel had four ceos.

the loss of mobile internet happened during paul otellini's tenure. in 2006, intel refused to manufacture the first generation of apple mobile phones. otellini's consideration at the time was not unreasonable: jobs pushed the bid to the extreme, far below intel's internal cost estimate. but unfortunately, this was not a cost calculation problem, but a vision measurement problem. intel missed the most important entry opportunity of mobile internet, and also "fatten" samsung.

brian krzanich took over as intel ceo in 2013 and served until 2018. during this period, intel was known as the "toothpaste factory". since breaking through the 14nm node in 2014, intel's foundry advanced process has encountered bottlenecks and gradually fallen behind tsmc.

as for the period from 2018 to 2021, that is, the early period before kissinger took over, bob swan once again made intel miss the opportunity to focus on mobile phones and gave up the research and development of 5g baseband chips. in addition, as mentioned above, during this period, apple also gave up intel.

therefore, when kissinger took office, he was highly expected. he joined intel at the age of 18, left for several years, and returned with the halo of a "technical person". this is a sinking ship. it takes courage to take over and it also takes courage to save the situation. either revive or sink.

it is not difficult to see that intel's seemingly reasonable but conservative choices at several previous opportunity nodes gradually deviated from its course. after taking office, kissinger immediately carried out drastic reforms and proposed the idm 2.0 strategy, which once reversed intel's decline.

it is an ambitious plan to make breakthroughs in advanced processes, take on external orders as a foundry, and compete head-on with tsmc and samsung.

ambition comes at a price, which means spending money on acquisitions and building infrastructure, but kissinger obviously does not want to repeat the mistakes of his predecessors who were "giants that were slow to turn around."

soon after taking office in 2021, kissinger began negotiations to acquire the global chip foundry globalfoundries, offering a price of $30 billion, but failed. not long after, kissinger wanted to acquire the chip foundry tower semiconductor for $5 billion, but also failed. at the same time, intel began to invest heavily in building wafer fabs, and established a separate wafer foundry services division in 2021 to open its wafer manufacturing capacity to external customers. intel's number of employees also reached a peak in 2022, exceeding 130,000 at one point.

intel also has an important external boost: the us chip act. it is called "the most significant intervention in industrial policy by the us government in decades", and the biden administration wants to realize the us "chip dream". as a domestic and established chip company in the united states, intel has become the "hope of the whole village", receiving a subsidy of us$8.5 billion (in fact, it has not yet been received, and the final allocation may also change as nvidia's situation changes).

in the semiconductor industry, catching up with competitors technologically requires huge investment and is bound to be a long-term battle. kissinger must have anticipated this.

however, despite all his calculations, kissinger did not anticipate the arrival of the ai ​​wave.

in fact, as early as the end of october 2022, a month before chatgpt came out, intel had already begun targeted layoffs and other adjustments, including reducing factory working hours, to cope with the economic downturn. at that time, intel did not disclose the number of layoffs, but claimed that it would increase annual cost reductions and efficiency gains to us$8 billion to us$10 billion by the end of 2025. in this way, intel will cut up to us$13 billion in three years.

on the one hand, kissinger's "intel recovery costs" continued to increase, and he had to take a few steps back. on the other hand, generative ai was booming. the market demand shifted from intel's cpu to nvidia's gpu. and intel wanted to catch up with the ai ​​wave, but compared with nvidia, which had an early layout, and amd, which responded quickly, it was still a little powerless.

as qualcomm and amd's ai chips are rushing to market, intel is also shouting that it is entering the "ai era." however, its first-generation core ultra cpu (codenamed meteor laker) has been slow to go into mass production, and its launch is nowhere in sight.

in august this year, kissinger said at the earnings conference that the ai ​​pc chip lunar lake is not enough to turn the situation around in a short period of time. intel must compete for the market share of server chips from other ai chip manufacturers. to succeed, intel's ai chips need to achieve comprehensive breakthroughs in performance, power consumption and even ecosystem construction.

it's just another big pancake, and it smells like helplessness.

ironically, just as people were surprised by the huge losses brought by intel's foundry business and lamented that it could not keep up with the ai ​​era, another piece of news in august this year made intel's image even worse - reuters quoted several people familiar with the matter as saying that intel had discussed investment matters with openai from 2017 to 2018. intel had the opportunity to acquire 15% of openai's shares for us$1 billion, and if it provided hardware to openai at cost price, it could acquire another 15% of the shares.

but almost as expected, intel eventually gave up the investment. at that time, openai was still a less well-known non-profit organization, and was quite interested in intel's investment, wanting to reduce its dependence on nvidia. in contrast, nvidia had established in-depth cooperation with openai early on. as early as 2016, nvidia ceo huang renxun personally moved the world's first dgx-1 supercomputer.

rather than saying that intel did something wrong, it is better to say that intel did not do the right thing and failed to successfully jump into the river of the times. little by little, it watched the river flow away from it.

intel is still actively trying to get out of its predicament.

after the release of the august financial report, in just over a month, kissinger said he would "focus on efficiency" and made a number of contraction measures. in addition to laying off 15,000 employees, he also suspended some factory construction investments in poland and germany: "we need to integrate the cost structure with innovative operating strategies to achieve a fundamental transformation in the way we operate. at present, our revenue growth has not met expectations, and we have not fully utilized the opportunities brought by cutting-edge technologies such as artificial intelligence. at the same time, our costs remain high, resulting in low profit margins."

in addition, intel has also won amazon's manufacturing contract and will expand its cooperation with aws to produce customized ai chips for it using the 18a manufacturing process (1.8nm process node).

as for the chips act, intel decided to cooperate with the us military and obtain us$3 billion from the biden administration to expand the supply of microelectronics products to the department of defense.

on the other hand, it must be emphasized that qualcomm's acquisition of intel has not yet been finalized, and it remains unknown whether intel can get out of the quagmire.

since the acquisition news came out on september 20, intel's stock price has risen, while qualcomm's stock price has fallen, and this trend continued after the auction started on september 23. by the evening of september 23, intel's market value had risen by about us$5.5 billion, while qualcomm's had fallen by about us$5.5 billion.

it cannot be ruled out that intel, which has accumulated many years of experience in chip manufacturing, will attract the attention of other ai manufacturers. whether it is technology giants such as microsoft and amazon, or star start-ups such as openai, they are actively seeking ways to get rid of or reduce their dependence on nvidia after accelerating its stock price.

the future of intel, which is currently in a state of illness, is uncertain, whether it will become nokia or usher in a turnaround.