Smart speakers are experiencing a major downturn. Can AI regain lost ground?
The smart speakers, once highly anticipated and predicted to be the gateway to smart homes, are facing a severe survival test.
Data shows that in 2024, the sales volume of smart speakers in the Chinese market was 15.7 million units, a year - on - year decrease of 25.6%, and the sales revenue was 4.2 billion yuan, a year - on - year decrease of 29.4%. From 2021 to 2024, the sales volume of smart speakers in the Chinese market has declined for four consecutive years.
Even with the "national subsidy", it is difficult to reverse the decline, and the decline rate still hovers around 20%. The once - prosperous smart speaker market attracted many leading Chinese Internet and home appliance manufacturers to enter, but now it has quietly cooled down.
Glory and Hidden Dangers Coexist
Before 2020, smart speakers were real stars in the market. The year - on - year sales growth rate of 125% in 2019 is strong evidence.
However, the good times did not last long. In 2020, the year - on - year sales growth rate of smart speakers dropped to 3.3%, and then it started a four - year journey of sales decline.
Historical experience shows that when an industry reaches its peak and then plunges rapidly, it is mostly due to "false prosperity".
During the early growth peak of the smart speaker market, due to the large number of entering enterprises eager to quickly seize the market, Tmall Genie, Xiaomi, and Xiaodu all resorted to the "subsidy strategy" to try to rapidly mature the market.
From 2018 to 2020, many domestic giants launched a fierce price war. The most extreme example was the Xiaodu Smart Screen, which dropped directly from 1,599 yuan to 299 yuan during the Double 11 shopping festival in 2018, a reduction of up to 1,300 yuan.
The price war ultimately brought two negative impacts. First, the affordable prices attracted a large number of consumers to buy smart speakers, but it also over - consumed the market potential, leading to premature market saturation.
Second, it led to the emergence of a batch of basic smart speakers priced at around 100 yuan, which were very similar in terms of function, appearance, and pricing, becoming a concentrated manifestation of the homogenization phenomenon of smart speakers.
There is no high - level technical barrier in smart speaker technology itself, so there will naturally not be much difference in the functional experience of products. Low - end products have average experience and poor compatibility, while high - end products, although having certain hardware advantages such as in sound quality, lack sufficient substitutability compared with professional speaker products.
Smart speakers themselves are also highly substitutable. Smartphones with extremely high market penetration can integrate more functions than smart speakers and are more suitable as the control center for smart homes.
Even without a smartphone, tablets and some home appliances can also perform the functions of smart speakers, which further reduces the advantage range of smart speakers.
Frequent privacy leaks have also invisibly accelerated the decline of smart speakers.
In 2022 in Benxi, Liaoning, Ms. Yang, who checked into a homestay, found that when playing with the smart speaker in the room, the built - in "care assistant" function had recorded and saved private videos of multiple guests, including six or seven videos of her and her friends walking in the room and a large number of private clips of previous guests.
As early as 2019, Bloomberg reported that Amazon hired thousands of employees globally to manually listen to and check the recorded conversations between users and the voice assistant Alexa captured by Echo smart speakers. These recordings would be transcribed and annotated to improve voice recognition technology. Although Amazon claimed that the recordings were not directly associated with user identities, employees might still hear users' private or even sensitive content, such as singing in the shower or sounds suspected of being related to crimes.
Technical limitations are also a reason for the decline of smart speakers.
Smart speakers were once called the "control center of smart homes" and could be used for remote control via mobile phones or voice control of smart home devices at home. However, in actual use, their technical limitations are very obvious.
The voice interaction experience has become the biggest pain point. Although voice assistant technology has been continuously improving, problems such as misrecognition, answering irrelevantly, and limited semantic understanding ability still exist commonly.
In 2018, Lei Jun had an embarrassing moment at the AIoT Developer Conference. When he asked "What character is formed by three 'Mu's?", Xiaomi's voice assistant Xiaoai did not give the correct answer "Sen", but unexpectedly started singing the pop song "Super Star". When he asked "What is the sum of 125 + 357 + 567?", Xiaoai simply stopped responding.
Of course, some explanations can be found. For example, in a noisy on - site environment, or when dealing with non - standard or slightly accented expressions, it is impossible to perform accurate recognition and understanding. Even the Bluetooth speaker priced at only 49 yuan, due to the hardware configuration such as the chip it uses, may indeed have some limitations in far - field sound pickup (especially at a press conference with echoes and noise), data processing speed, and response ability.
However, similar scenarios may also occur in daily life, which leads users to often feel frustration (frustration caused by product defects) during use, especially in complex scenarios, where the response speed and accuracy of smart speakers are difficult to meet the needs.
The weakness of the content ecosystem also restricts the development of smart speakers. High - quality content resources such as music and audiobooks often require additional paid subscriptions, and the resource repetition rate is high, making it difficult to meet users' diverse needs.
The Tripartite Confrontation and Market Shrinkage
The once - prosperous smart speaker market attracted many leading Chinese Internet and home appliance manufacturers to enter. However, due to the lack of functions of smart speakers, consumers lack the motivation to upgrade. The number of brands with detectable sales in the market has decreased to 11, 8 fewer than in the first quarter of last year.
Enterprises over - consumed the market potential, and devices such as tablets, mobile phones, and smart TVs have impacted the position of smart speakers, leading to the continuous shrinkage of the smart speaker market. Small brands are withdrawing, and the market is becoming more and more concentrated among the leading players.
Today, the combined market share of Xiaomi, Baidu (Xiaodu), and Tmall Genie has always remained above 90%. In the first quarter of this year, their combined share even reached 96.5%.
Even so, the leading players still have to face the difficult reality of continuous sales decline. The emerging AI has become the life - saving straw for smart speakers.
The way smart speakers are used, which is to control smart home devices at home, answer questions, or simply communicate according to users' voice commands, seems to be a perfect fit for AI large - scale models.
Leading Chinese smart speaker manufacturers are indeed trying to integrate AI with smart speakers. For example, Xiaomi has fully pushed the Xiaoai version based on the large - scale model to multiple devices, and some devices will be updated successively within October.
Baidu and Alibaba, which respectively own the Wenxin and Tongyi large - scale models, naturally won't miss the AI wave. For example, the Xiaodu Smart Speaker MatePro under Baidu, based on the Wenxin large - scale model and the DUER OS system, can achieve AI casual Q&A, accompany users in chatting, and can also recognize dialects.
Currently, new products on the market are all equipped with AI large - scale model technology. In the first quarter of 2025, the market penetration rate of devices supporting AI large - scale models exceeded 20%.
However, the problem is that the integration of AI large - scale models has not changed the essence of the smart speaker industry, that is, the ecological problem.
In the past few years, the most important function of smart speakers, which evolved from simple conversation, is to control smart home devices at home. The addition of AI large - scale models makes smart speakers smarter and more accurate in understanding user instructions, but it cannot enrich the smart home ecosystem.
One of the biggest problems in the smart home ecosystem at present is fragmentation. Each giant hopes to build its own ecosystem, resulting in difficulties in seamless cooperation between devices of different brands.
If you have a smart light from Xiaomi's Mi Home, a smart socket supported by Alibaba's Tmall Genie, and a smart camera under Google's Google Home ecosystem at home, you may need to download three different apps (Mi Home, Tmall Genie, Google Home) to set up and control these devices respectively.
Although some platforms try to achieve interoperability through technical cooperation or agreements (such as Matter), deep - level functions and automated scenarios often still cannot be used smoothly across ecosystems. For example, you may not be able to directly call up the picture from a Google camera using Tmall Genie, or let Xiaomi sensors accurately trigger devices in the Alibaba ecosystem to perform complex tasks.
This ecological fragmentation forces consumers to "take sides" when making purchases or endure the hassle of operating multiple apps, which is contrary to the original intention of a "seamless smart experience".
The high - level cloud dependence of many smart devices is the root cause of their vulnerability. Once the manufacturer's server has problems or stops service, the functions of the devices will be severely damaged or even completely fail.
The smart lighting brand Sengled was removed from the "Works With Alexa" project by Amazon because its server continuously had downtime problems that could not be solved, resulting in users being unable to control its smart bulbs through the Alexa voice assistant.
Even within the same ecosystem, technical failures and poor experiences can seriously undermine users' confidence in smart homes.
In March 2025, some users reported that Xiaoai could not control smart home devices, and there were also problems such as data not being able to load in the Xiaomi Speaker app. Xiaomi's technical team had to conduct an emergency investigation and repair. This failure directly left users' smart devices at home in a semi - paralyzed state, highlighting the vulnerability of the centralized control node.
What's the Fate?
Before smart speakers can regain new vitality, potential substitutes have quietly emerged.
At conferences such as AWE and CES held this year, there were many AI toys developed by Chinese manufacturers. Some of these products focus on the companionship function and are connected to large - scale models such as DeepSeek and Tongyi Qianwen.
These on - sale AI toy products are quite similar to smart speakers with AI capabilities in terms of functions. The difference is that AI toys focus more on communication and can serve as companion robots or educational aids for children.
Besides route transformation, what smart speakers most need to do is to upgrade the software and hardware ecological system. At the hardware level, it means to cooperate with more home appliance manufacturers to increase the number of home appliance products connected to the ecosystem; at the software level, it means to optimize the AI experience of smart speakers.
Lotu Technology does not think that the national subsidy can help the smart speaker industry reverse the sales decline trend. It predicts that the sales volume of smart speakers in China in 2025 will be 13.5 million units, a year - on - year decrease of 14%.
The smart speaker industry seems to have reached a dead end. However, the in - depth integration of AI technology may bring a glimmer of hope.
In February 2025, the Chinese smart speaker market witnessed its first significant rebound. The online sales volume reached 357,000 units, a year - on - year increase of 12.0%, and the sales revenue reached 94 million yuan, a year - on - year increase of 10.0%.
This growth trend broke the consecutive - year decline situation, demonstrating the continuous breakthrough and innovation ability of AI technology in the field of smart homes.
For industry practitioners and consumers, understanding and grasping this wave of AI technological innovation trends will help them take the initiative in the industry's development and also provide a solid technical foundation for the intelligent upgrade of smart life.
Smart speakers are not completely a failure, but they need to re - find their positioning.
This article is from the WeChat public account "Decoding NewSight", author: Shi Gaofei, published by 36Kr with permission.