The AI space is fiercely competitive, and Google once faltered in the generative AI race, but this year’s I/O conference showcased many achievements, demonstrating its stability and strong position. This article analyzes the organizational response, full-stack thinking, and long-term accumulation behind Google’s success, as well as Baidu’s similar accumulation and strategic patience in the AI field, showcasing the competitive landscape of the two companies in the new stage of AI.
The competition in the field of AI is far from the moment when Ming Jin closes his troops.
In the past week, from OpenAI’s acquisition of Apple’s former chief design officer Jony Ive in an all-stock transaction at a valuation of $6.5 billion, to Anthropic’s release of Claude 4, which takes a step forward in programming capabilities, the industry has remained intensive. However, even in the midst of the noisy updates, the Google I/O developer conference is still the one with the most “stamina”. In the circle of AI, people are still trying to understand the changes that are taking place in Google and the entire AI wave through this year’s new I/O conference.
Over the past year or so, Google seems to have faltered in the race for generative AI. However, this year’s I/O conference, whether it is the all-round upgrade of the Gemini 2.5 series, another generational iteration of Veo 3 sound and picture synchronization, or the official entry of Project Astra into the real product, serving all users in the form of Gemini Live, and the “AI Mode” announcing that AI will fully take over search, and the penetration of the ambitious AI Agent model in the entire product line, all clearly send a signal: Google has not only stabilized its position, It has also begun to occupy a favorable position in the new competitive stage.
The reason why Google I/O has sparked ongoing discussion is far more than the power of these products themselves, but also because it happens in a delicate moment of “potential energy transition”. Why can Google, which was once considered to be “getting up early and catching up with a late set”, be able to turn the situation around in a short period of time?
B-end product manager’s ability model and learning improvement
The first challenge faced by B-end product managers is how to correctly analyze and diagnose business problems. This is also the most difficult part, product design knowledge is basically not helpful for this part of the work, if you want to do a good job in business analysis and diagnosis, you must have a solid …
View details >
One immediate reason is the quick response and focus of the entire organization. A noteworthy signal is that compared with the executive lineup sitting in front of the Google I/O meeting last year, the Google executives sitting in front of the silicon star this year have a clear division of responsibilities, and the matters responsible for the main business and technical departments have been sorted out more internally. The deeper reason behind this is rooted in Google’s full-stack thinking and long-term accumulation of “old books” in AI.
As Google CEO Sundar Pichai mentioned in an exchange with Silicon Stars, the research and development of TPUs began nearly two decades ago, and Waymo’s accumulation is not a one-day achievement. This full-link control from chips, models, platforms to applications, and continued investment in technology and the company’s understanding of technology during low periods have enabled Google to quickly mobilize resources and optimize the efficiency of its entire technology stack today.
This reversal of potential energy stems from a deep understanding of the core business and corresponding technologies over a longer period of time. Google proves two things with its actions:
First, no matter how iterative AI search is, its foundation is still inseparable from the massive data, user understanding, and crucial infrastructure accumulated by traditional search services. New AI experiences, such as AI Mode, essentially superimpose a layer of intelligent summarization and guidance powered by large models on top of existing search results.
Second, no matter how strong the “sense of generation” brought by large models is, it is still part of the development process of AI technology. The accumulation in the fields of natural language processing, machine learning, and knowledge graph in the past has not been written off, but has become the cornerstone of rapid iteration and performance improvement of new models.
The essence of the search business is to transform the algorithms in the laboratory into products that meet the real needs of hundreds of millions of users. Google has been going through this process for more than 20 years. From the original PageRank to later search, to BERT’s revolutionary improvement in search understanding, to today’s Gemini. Google knows how to engineer, productize, and build sustainable business models around cutting-edge algorithms. This experience allows Google to better grasp the rhythm of moving from technological maturity to practical application and promote the continuous rotation of the innovation flywheel in the face of the new technological wave of large models.
Google’s return to the top has made many people in the AI industry curious about who can match Google’s full-stack and long-term accumulation capabilities in the next competition. And obviously, such a company needs to be the same as Google, which has been there since the beginning of the story and has been at the poker table to this day.
In the exchange with the Google search team, they mentioned that the process of Page Rank technology eventually becoming a valuable application of search, which also brings strong reference to the decision-making process of today’s Google model. Another company that completed the same process in search during the same period was Baidu.
More than a decade ago, the direct competition between Google and Baidu, although in the name of search, actually opened the beginning of today’s AI – search was one or even the only business that required a large number of algorithmic talents at that time, and AI talents completed the initial queue and division at that time. The well-known auction between Baidu and Google for Geoffrey Hinton took place under this logic. After that, Baidu and Google also launched a catch-up race in machine translation, a key application of AI in the early days. Baidu once released machine translation papers continuously, showing that it was comparable to or even surpassed Google’s internal performance, and Google’s internal anxiety was finally alleviated by the timely application of TPU, and it released a powerful translation service before Baidu.
After that, the two companies also laid out autonomous driving at the same time.
In the field of autonomous driving, Google’s Waymo and Baidu’s Apollo are almost the earliest and most resolute players in the world, and today they have finally become the only two platforms for large-scale operation. As Sundar Pichai has emphasized many times before, even in a period when the autonomous driving industry is facing many doubts and is generally looked down upon by the outside world, Google still chooses to continue to increase investment in Waymo.
Baidu’s investment in Apollo and Radish Run has also lasted for more than ten years, and has experienced a complete cycle of the industry from fanaticism to calmness to gradual commercialization. Both sides deeply understand that autonomous driving is the crown jewel of AI technology, and its complexity and extreme requirements for safety determine that this must be a “hardcore game” that requires long-term and continuous investment. Whether it is Waymo’s continuous investment in sensor fusion and simulation test platforms, or Baidu’s Apollo’s exploration of vehicle-road collaboration and L4-level autonomous driving, they all reflect the same technical strategic thinking: that is, to carry out full-stack layout in key technology fields, and overcome technical difficulties with great patience and resources, and finally promote the maturity and commercialization of technology.
These intriguing similarities in core technology genes and strategic cognition of AI make people look forward to Baidu again today.
Baidu and Google started from a similar starting point (search) and also went through the process of transforming algorithms into large-scale applications. Moreover, Baidu is not just a follower. In the long development of AI, Baidu has almost never been absent from key nodes: from the early investment in deep learning, to the research and development of voice and image technology, to the decade of deep cultivation of autonomous driving, and the all-out efforts to large models in recent years. Baidu has always been sensitive to the evolution of AI technology priorities and continues to apply algorithm understanding and engineering experience derived from search and other businesses to actual products.
This accumulation, as well as the unique cognition formed in processing massive data and understanding user intentions, is also an opportunity for Baidu in the second half of the large model today.
“I don’t think innovation can be planned, you don’t know when it’s coming, all you can do is create an environment conducive to innovation.” Robin Li, founder and CEO of Baidu, who has personally experienced the entire wave of AI all the way, once expressed his views on today’s technological anxiety. To some extent, this also reveals Baidu’s strategic patience in the field of AI. He emphasized, “When technology is developing so fast, you have to keep investing to ensure you are at the forefront of technological innovation. We still need to continue to invest in chips, data centers, and cloud infrastructure to train better, smarter next-generation models. ”
Baidu’s “four-layer AI architecture” includes a cloud infrastructure layer with a 10,000-card cluster, a flying paddle open source framework layer widely used by Chinese developers, a constantly iterative Wenxin large model layer, and an application layer such as Baidu Search and Baidu Library for AI reconstruction.
Recently, Baidu announced at the Create conference in April 2025 that it had lit up the first fully self-developed 30,000-card cluster in China, which provides a solid computing power foundation for its large model training and inference. In the process of transforming AI technology into practical applications and business value, cloud platforms play a crucial role. For Baidu, Baidu Intelligent Cloud is not only a window for its AI technology output, but also one of the core engines of its AI commercialization strategy.
According to Baidu’s financial report for the first quarter of 2025, cloud revenue increased by 42% year-on-year, driven by AI, and the AI-contributed revenue related to cloud business has reached triple-digit growth, and the operating profit margin has also exceeded 10%. According to statistics, 65% of central state-owned enterprises are currently using Baidu Intelligent Cloud.
At the model level, the Wenxin model is iterating rapidly. From Wenxin 4.5 Turbo to the deep thinking model X1 Turbo, Baidu not only emphasizes multimodal processing and strong logical reasoning capabilities, but also continues to optimize costs.
Wenxin Model 4.5 Turbo is faster and the price is reduced by 80% compared to Wenxin 4.5, and Wenxin Model X1 Turbo has improved performance and reduced price by another 50% compared to Wenxin X1. The average daily call volume of the Wenxin model exceeded 1.65 billion, and the number of users of Wenxin Yiyan reached 430 million.
“We live in a very exciting time. According to Moore’s Law, every 18 months, performance doubles and price halves. But today, when we talk about large language models, inference costs can basically be reduced by more than 90% in 12 months. Robin Li said at the recent Baidu earnings conference.
“Not only in the AI field or the IT industry, looking back at the history of the past few hundred years, most innovations are related to cost reduction. If costs are reduced – proportionally, productivity will increase by the same proportion, which is the essence of innovation. Today, innovation is much faster than ever. ”
This is also Baidu’s thinking on ecological construction. Baidu Intelligent Cloud’s Qianfan large model platform plays a key role. The platform has been connected to hundreds of mainstream large models at home and abroad, providing developers with a wealth of model choices and highly competitive prices. At present, Qianfan has helped customers fine-tune 33,000 models and develop 770,000 enterprise applications. It is also the first cloud vendor in China to be compatible with MCP. It is hoped that the sharing and invocation of AI capabilities will be facilitated through standardized interfaces.
In the first quarter of 2025, Baidu Intelligent Cloud ranked first in the bidding market of general large model manufacturers with 19 winning projects and a winning bid amount of 450 million yuan.
Moreover, in some specific AI product concepts and landing rhythms, Baidu has begun to show a more sensitive sense of smell. Taking agents as an example, Robin Li regards it as “the hottest track for AI applications”. To this end, Baidu has launched the universal “super agent” product “Xinxiang” App, and the no-code generative application development platform “Miaoda”, aiming to lower the threshold for the development and use of AI applications.
An interesting detail is that in the design of the product “Wen Xiaoyan”, Baidu has automatically selected the most suitable model to handle specific tasks according to user needs. According to the previous communication between Silicon Star and the Google Gemini team, Gemini has initially realized the automatic call of different capabilities of the same model according to user needs, and the longer-term goal is also the automatic selection of different models. To a certain extent, this reflects Baidu’s pre-thinking on the combination of user experience and technology in specific application scenarios, and even has a bit of the meaning of machine translation.
In the field of autonomous driving, “Radish Run” has begun to expand to the international market, and Baidu announced that it has reached a strategic cooperation with the Dubai Road Transport Authority (RTA) and Autogo in the United Arab Emirates, planning to provide driverless travel services in Dubai and Abu Dhabi. Its sixth-generation unmanned vehicle costs only one-seventh the price of Google’s Waymo model. In addition, the world’s first L4 end-to-end autonomous driving model Apollo ADFM released by Baidu Apollo is also beginning to explore the next technical node.
Today, AI is gradually moving from the show-off skills of a single model to a deeper level of ecological construction and value implementation. Although the simple chatbot form has attracted a lot of attention in the early stage, it has shown limitations in terms of user retention and business model sustainability. In contrast, those companies committed to building “full-stack services” show greater resilience and development potential. From the comprehensive AI reconstruction application “Family Bucket” displayed by Google I/O to the simultaneous flowering of Baidu’s four-layer structure – especially on the application side, the old tree is blooming with new flowers of AI, Baidu Wenku AI monthly active users reach 97 million, Baidu Wangpan monthly active users exceed 200 million, AI monthly active users exceed 80 million, and the average daily storage of files exceeds 1 billion.
Looking at today’s competition in the long history of AI technology development, we will find that it has always been a process of succession and alternating leading. The final test is endurance and vision, as well as the accumulation of various “persistences” along the way, Baidu and Google’s “paranoia” of technology has not been fully recognized by the outside world for a period of time when ChatGPT brought FOMO, and today when everyone realizes the importance of long-term technology accumulation, full-stack strategic layout and long-term adherence to core business, the potential of those companies that are closest to this technology, the deepest accumulation, and the deepest understanding is finally reseen.
