Manus challenges OpenAi and fires in the Chinese technology race

Deepseek and manuses challenge western rule in artificial intelligence, but American restrictions can put a brake on this rise

The wave of enthusiasm around Artificial Intelligence (AI) in China has gained a new breath with the release of Bot Manus, developed by Butterfly Effect. Within hours after its release on March 6, the registration platform was overturned by a flood of access. The company states that its technology exceeds the models created by OpenAi, responsible for ChatgPT. At the moment, access to service is restricted and requires invitation, while the company deals with excess demand. There are reports of resellers selling access codes in the black market.

The manus is just the most recent example of a phenomenon that has taken over the country since January, when DeepSeek, a Chinese AI startup, surprised the world by presenting a low -cost advanced model. This innovation has had a significant impact on local markets.

Chinese actions have registered the best start of the year in history, with Hang Seng Tech Index, which follows the largest technology companies listed in Hong Kong, jumping over 40% since mid -January.

Many experts believe that the reduction of IA costs will pave the way for new technological applications. Cloud computing providers are expanding their investments on data centers, boosting spending throughout the production chain. However, doubts arise about what could interrupt this growth cycle.

In recent weeks, hundreds of large Chinese corporations have announced plans to integrate Deepseek technology into their business. These include automotive automakers, state -owned power companies, banks and food and beverage manufacturers. Giants like Tencent are also adopting the solution, even having its own AI models.

Municipal governments are incorporating Deepseek systems into applications used by citizens to access public services, while government departments, hospitals and universities discuss ways to use it to “strengthen the communist party.”

Local analysts joke that any report needs to mention Deepseek to attract attention. Investors speculate that startup could, alone, revitalize the Hangzhou real estate market, where it is headquartered.

Chinese risk capital investors are ecstatic. A Beijing investor celebrates the results of integrating Deepseek technology in portfolio robotics companies, highlighting cost reductions and performance improvements. In this scenario, AI startups are emerging in mass all over the country. Some investors are injecting capital into these companies, even recognizing bubble signs.

“It’s overwhelming, but we have no other choice,” says a Hangzhou investor. “The economy is not good and there are few opportunities in other sectors. So we need to get in the AI ​​as soon as possible. ” The strategy is to invest in an initial financing round, known as “A” round, and leave during a “A+” round, which can occur months later. On March 6, the Chinese central government announced the creation of a 1 trillion yuan fund ($ 140 billion) for technology investments.

China’s leading technology companies, including Alibaba, Baidu, Huawei and Tencent, are taking advantage of the wave and hopes to profit from boom through their cloud computing divisions. Last month, Alibaba stated that its main objective was to achieve general artificial intelligence similar to human. On March 6, the company launched a new reasoning model that claims to be as efficient as DeepSeek’s.

Alibaba has promised to invest about $ 53 billion over the next three years in building data centers to meet the growing demand for AI cloud services, surpassing their investments from the past ten years. The company leads the cloud market in China, with 36% share, and bets that growth in this area will compensate for the slowdown of its e -commerce business.

Baidu has seen an increase in the revenue of its cloud division, helping to compensate for falls in other areas. AI demand can also improve profit margins in the Chinese cloud computing sector, historically smaller than in the West due to strong competition.

According to The Economist magazine, the demand for Specialized AI servers has fired since the end of the Chinese Lunar New Year in early February, coinciding with the rise of DeepSek.

Suppliers have begun to offer pre-equipped “everything-in-one” servers with AI software. Many of these equipment is sold directly to companies that prefer to keep them in their facilities to increase safety, including state-owned companies.

Sangfor Technologies, founded by former Huawei employees, was one of the biggest beneficiaries: its shares rose about 140% this year. Analysts estimate that the server market all-in-one will grow more than 70% per year, on average by 2028.

AI boom is boosting investments throughout the hardware supply chain. Investment Bank Jefferies estimates that server manufacturers can spend more than 1.4 trillion yuans over the next two years to expand its productive capacity.

GDS, one of the largest in the sector, has increased its capital spending plans. Already VNet, its competitor, announced that it will double its capacity this year.

However, some analysts warn of the need for caution. Morningstar’s Kai Wang argues that DeepSeek will not change the foundations of most companies that have benefited from the recent Chinese market rally. An earlier Rally lost strength when government support to the economy did not materialize; Something similar may occur this year, says Wang, if companies face difficulties to monetize AI.

Advanced semiconductor access can also be an obstacle. Currently, companies can purchase H20 chips from Nvidia, the American AI chips leader.

Although these chips are less powerful than the most advanced Nvidia models, whose sale to China was banned by the US, they have worked well. Local chip designers such as Cambricon, Enclame and Huawei are trying to reach competition and already provide for some Chinese startups.

Even so, semiconductor scarcity can curb the enthusiasm around AI in the country. As new applications arise, increasing the demand for computational power, restrictions on chip offering can start to weigh. SMIC, the main Chinese state foundry, faces serious capacity limitations and cannot produce the most advanced semiconductors.

In addition, the best huawei designed chips are still far behind NVIDIA in terms of performance. Greg Allen of CSIS, Think Tank based in Washington, recently wrote that he will take years for Huawei AI chips and associated software reach a competitive level.

The Trump government is said to be considering more severe measures against China, including restrictions on access to H20 chips. The current Chinese rally is based on the expectation that the costs of training and operating AI models will continue to fall.

By restricting access to advanced chips, the US president could increase these costs dramatically, putting an abrupt end to the euphoria around AI in the country.