Author | Xue Xingxing Editor | Jiang Jiao

Cover | "Silicon Valley"

On July 7, Baidu held a director's meeting internally. This communication meeting is called "Fireside Chat" within Baidu, held quarterly, attended by Li Yanhong and management personnel at the director level and above within Baidu, and not open to grassroots employees. Baidu released the minutes of the director's meeting internally on July 12.

A document exclusively obtained by "Shangshan" shows that in addition to Li Yanhong's oft-repeated AI strategic issues such as closed-source large models being better and agents being the direction of AI application, he rarely responded to Baidu's current governance challenges at this meeting, including low morale within the company, senior executives not taking responsibility, and difficulties in cross-departmental collaboration.

Two months ago, Qu Jing, the head of Baidu's public relations, resigned due to several controversial short videos, sparking questions about Baidu's culture both inside and outside the company. This Fireside Chat was held for the first time after Qu Jing's resignation. The conversation was between Liang Zhixiang, Qu Jing's former supervisor and the current senior vice president in charge of Baidu's public relations business, and Li Yanhong.

Li Yanhong did not mention Qu Jing at the meeting. When answering a question about what makes an excellent middle and senior manager, he said that managers must have leadership and set an example for their subordinates. "Many times, culture is not written down on paper, it's not something you can say or write, but rather how you solve problems when they arise."

He said that managers who don't understand the industry and can't make decisions shouldn't be in those positions. At Baidu's Advanced Employee Awards ceremony in early May, Li Yanhong said that he considers himself a relatively good-tempered boss who rarely gets angry with employees, but he also has a bottom line that must not be crossed.

In Li Yanhong's latest quarterly OKRs, O4 mentions timely alignment of organization and strategy, KR1 mentions learning AI-native thinking to further focus on the company's strategic direction, and KR2 mentions timely summing up of experiences and lessons and aligning cross-organizational communication. Li Yanhong's explanation of the OKRs did not involve the Qu Jing incident, saying that all these OKRs were set based on AI direction.

01 | "This time will be different"

In the first half of 2023, riding on the wave of large models, Baidu's share price continued to rise, once soaring to HK$154.4 in March that year, reaching its highest point since early 2022. However, it has since continued to decline, erasing all gains. Even though it recently hit its highest single-day gain in over a year due to the rise of Luobo Kuaipao, it has not returned to last year's high. Over the past year, Baidu's Hong Kong share price has fallen by 37%.

Some market analysts believe that Baidu's current AI strategy has not yet significantly driven its main business, and online advertising remains the backbone of Baidu's revenue. This business has seen slowing growth for several consecutive quarters, and there are still no signs of recovery in the second quarter. Li Yanhong said on Baidu's first-quarter earnings call that Baidu has been using generative AI for advertising business commercialization since the second half of last year, currently bringing in millions of yuan in incremental revenue to the advertising business each quarter.

At the internal director's meeting, Li Yanhong also rarely admitted that Baidu's growth rate has come down, "it's a difficult time." He hoped that employees would not pay attention to external opinions and focus on delivering results, and other things would come naturally. "If you believe in this, stay, let's do it together. If you don't believe in it, there are better options outside, that's it."

Li Yanhong said that the current state of the mobile internet has become a zero-sum game, and Baidu is in a passive position. "There is still a big gap in the content ecosystem compared to Douyin and WeChat. If we want to catch up by following the mobile internet playbook again, it's not that easy."

He said that Baidu's morale is indeed not good at the moment, which is due to Baidu being in a transition process. Past technological breakthroughs have led them to mistakenly believe that the era of AI has finally arrived, but "each time, the best outcome is quite disappointing." He believes that "this time will be different." It remains to be seen how long it will take to usher in the true era of artificial intelligence.

Li Yanhong believes that the anxiety and sense of urgency currently felt by employees are normal. He also cited the example of Li Xiang's MEGA, "We can see that these people or companies have a stronger sense of anxiety than we do." Li Xiang gave a speech in the company corridor in 2022, saying that they had already been targeted by the world's most predatory companies. But even so, situations like MEGA not selling well still occur, so a sense of urgency is necessary.

The solution he gave is "focus." Li Yanhong said that Baidu can't fight too many battles, and the company's resources are limited. They need to put resources where they truly relate to core business, avoiding excessive resource dispersion that could lead to losing battles that should have been won.

A former IT employee at Baidu told "Shangshan" that Baidu used to be a company that was particularly willing to engage in new businesses, "because it didn't have any particularly outstanding businesses. To put it bluntly, if I independently started a project, it would take off, but basically, none of them did."

Over the past decade, Baidu has tried many cross-border businesses in addition to AI and search, including takeout, group buying, e-commerce, live streaming, short videos, and more. Before its return to Hong Kong listing, it had planned to spend US$3.6 billion to acquire YY to make up for shortcomings in its mobile ecosystem. This acquisition, which was Baidu's largest in history, was terminated after several years of delay.

Before the large model, its most recent abandoned strategic direction was the metaverse. In 2020, Baidu launched China's first metaverse product, "Xirang," and even held the next year's Baidu AI Developer Conference within Xirang. However, Xirang performed poorly, with graphics even inferior to mobile games from a few years ago. Last year, Xirang's head, Ma Jie, announced his resignation, which was seen by outsiders as a sign that Baidu was abandoning the metaverse.

Baidu Xirang

"One startup project after another, none of them have been particularly useful, basically none of them have taken off, and in the end, they all failed." The former Baidu IT employee said.

Li Yanhong believes that it's normal that they lose many battles because many of the things they do have not been done by previous generations, and they go to war without knowing whether this path is viable or not. "After three to five years, we may find that this path is not viable, which is not uncommon, and I accept it because no one in the world has walked this path before." They are at the forefront, and they must acknowledge this rule. They need to achieve nine failures and one victory.

02 | Management must have leadership and set an example for their subordinates

Baidu's core values are "Simple and Reliable," which have been used for over 20 years, but it seems that they have not yet deeply imprinted the organizational culture onto every employee like Alibaba. A Baidu employee told "Shangshan" that they don't understand these words very well. Another former employee said that internal employees even have a sarcastic saying about it, "Simple, can rely (on others)."

The Qu Jing incident in May further intensified questions about values both inside and outside the company. In early May this year, Qu Jing, the head of Baidu's public relations, posted several controversial videos on short video platforms, attracting public attention. Qu Jing later apologized on her personal social media account and soon resigned.

Cui Shanshan, Baidu's senior vice president, said at the company's internal Advanced Employee Awards ceremony that it's not strange for the company, which is almost 25 years old, to have problems, but it's wrong not to view them soberly and rationally. She said, "At Baidu, nothing is more worth adhering to and defending than the values of Simple and Reliable." Li Yanhong also said at the event that these outstanding employees being honored are the truest representatives of Baidu.

Li Yanhong at Baidu's internal awards ceremony / Image source: Baidu

Several employees interviewed by "Shangshan" mentioned management problems within Baidu, such as middle-level leaders not taking action, passing the buck, and only following orders from above, "I'll do whatever the top wants me to do."

One of the questions at the director's meeting was about what the criteria for excellence are for middle and senior managers at Baidu. "There are many high T, high P, and even M4, M5 levels in the company, 'sitting in the temple, unaware of the distant world,' not taking responsibility at critical decision points, requiring business departments to do a lot of internal educational work, leading to low efficiency, long processes, and slow decision-making. Robin has also said at Baidu's Pride Awards ceremony that these outstanding employees represent Baidu's cultural values. So, what are the criteria for excellence for middle and senior managers?"

Li Yanhong replied that the company has a history of over 20 years, and some of those directors or M4, M5 levels, including T10, P10 and above, have indeed gradually become detached from specific work. He said he has requirements for HR to assess leaders, and if a leader doesn't understand the industry, can't make decisions, they shouldn't be in that position. For executives, their requirement is that "whoever benefits should drive it forward, and if it can't be solved, escalate it (rise up the hierarchy)."

"Both leaders and employees must understand that everything will have a conclusion and a solution. If no one can solve it in the end, it will rise to me, and I will definitely give you an answer and a decision. It's impossible for anything to drag on indefinitely without resolution. So, don't let this delay our decision-making and execution," Li Yanhong said.

Li Yanhong said that in addition to the requirement of "Simple and Reliable," he also has a requirement for leadership from managers. "Your subordinates look up to you, so you need to set an example for them." He addressed the middle and upper-level employees at Baidu's director level and above, "In the position of Baidu's directors, everyone must be able to set such an example, letting more people feel that this is a company full of positive energy, competitiveness, and promising future."

"Baidu used to be a bit like a state-owned enterprise," the aforementioned Baidu IT employee told "Shangshan," but now Baidu has started to reduce costs and increase efficiency, with strict attendance checks. A Baidu employee said that the company's punch-in time is tracked, and employees are expected to arrive at work at 9 am but usually leave around 8-9 am.

A former Baidu employee who was laid off recently said that at first, he didn't know if his layoff was the intention of HR or department leaders. Later, several laid-off employees analyzed it privately and believed it was the AI's decision, "calculating your off-work time, calculating your code submissions, calculating your overall cost-effectiveness."

Layoffs have been ongoing intermittently. According to Baidu's annual report, the total number of Baidu employees at the end of 2023 was 39,800, compared to 41,300 at the end of 2022, a reduction of 1,500 employees in one year. However, this figure is still smaller than the number of layoffs at Alibaba and Tencent.

The aforementioned former Baidu IT employee said that in the past few years, Baidu's R&D positions at T6 and T7 could directly transition to management and start at the M level, where salaries were higher and easier to progress, resulting in no one writing code. But in the current context of cost reduction and efficiency enhancement, "they started to cut these management layers, especially those who don't write code." In the past, Baidu employees had an annual salary adjustment of around 6%, which has now been abolished.

03 | If you don't speak up, you'll be misled

The past period may have been the most intensive time for Li Yanhong to speak out externally. He constantly talked about AI strategy, sparking many controversies, such as "closed-source large models are definitely better than open-source ones." His most recent view is that agents are the promising direction of AI application, and search is the largest entry point for agent distribution.

Li Yanhong said that he was the first guest speaker at the industrial forum of the Shanghai Artificial Intelligence Conference, and before speaking, he anticipated that many of his statements would cause controversy. But the current AI market is chaotic, "people from all walks of life are expressing various opinions on AI, and if we don't speak up, we'll be misled by others," so he wants to continue to output externally, trying to influence external understanding of the current state of AI development and future development directions.

Li Yanhong at the Creat 2024 Baidu AI Developer Conference / Image source: Baidu

Over the past year, almost all of Baidu's businesses have been integrated with AI. At Baidu World last October, Li Yanhong announced that all of Baidu's applications have been reconstructed through AI, including Baidu Search, Baidu Library, Baidu Netdisk, etc.

The aforementioned former Baidu IT employee said that the current employee promotions and business priorities will consider the relationship with AI. Some business lines that were originally far from AI also have to think about how to integrate AI, "whoever is related to AI when it comes to promotion, their promotion will be faster, it's that simple."

At this director's meeting, Li Yanhong reiterated his views on AI strategy, in addition to his previous external discussions on closed-source being better than open-source, agents, large models having broader prospects in ToB than ToC, but also including multimodal, Apple Intelligence, large model price wars, and more.

Multimodal is the right direction, EB5 will be compatible with multimodal

"I believe GPT-4o is the right direction. Before that, everyone was looking forward to the release of GPT-5. But after GPT-4 was released, we saw GPT-4V, GPT-4Turbo, GPT-4o, which are all various attempts based on GPT-4. The 4o direction is multimodal, with faster response speed, more natural connections, and more native, able to process various images, sounds, texts, etc., in one model, enabling better user experience and application in more fields. I still quite recognize the 4o development direction."

Li Yanhong said that they have also heard some voices on multiple business needs, and no matter whether they will release EB5 or any new model in the future, they will ultimately be compatible with multimodal.

Apple Intelligence has little to do with Baidu

Apple Intelligence has little to do with Baidu, as it mainly emphasizes large models on the device side, but Baidu has little control over the device side. Domestic companies like Huawei, Xiaomi, Oppo, and Vivo, as well as foreign companies like Apple, are very dominant in channel management. They may see the device-side model as an opportunity or at least something they can't miss. As for what they will eventually achieve, honestly, I don't care that much, it has little to do with us.

Robin Li believes that these companies might perform poorly with true cloud-based models, and it's likely they will rely on one or more cloud-based large models in the future. Baidu also hopes to provide them with the capability to use cloud-based models.

"As for future innovations with edge-side models, since I don’t pay much attention to them, I haven’t thought much about it. If it's merely copying the capabilities of cloud-based models, then there's nothing novel about that—it just ends up being smaller and less capable. At most, people might think that privacy protection is better since the data doesn't get transmitted to the cloud. That's about it."

Baidu doesn't compete on price; it ultimately competes on efficiency

According to Robin Li, the current market for large models is quite chaotic. There isn't a strong consensus on which model is the best, which offers the best cost-performance ratio, or whether to use public cloud or private deployment. They need some more time to move forward, and in the end, the competition will not be about price but about efficiency.

"Whoever achieves the highest efficiency—delivering the same results at the lowest cost or the best results at the same cost—will emerge as the winner in future competition."