Internet giant: With AI quality inspection as the “sharp knife”, cutting into the trillion-scale industrial Internet market

Different from Bohan Intelligent taking the Silicon Valley elite technology route, Tencent, as a product-driven Internet giant famous, sees a market of 100 billion. At the same time, Tencent strives to treat AI quality inspection as a “sharp-knife” product to cut into the next trillion-scale industrial Internet market.

In the dialogue with all media reporters of Southern Finance, Huang Qiang, product director of industrial AI of Tencent Cloud, said that as a rigid demand link in the industrial scene, AI quality inspection is just like a box of eggs in the community group purchase. Using this as a drainage tool, it becomes the first industrial APP for industrial customers to contact Tencent Cloud. After the connection is established, Then explore other digital factory number intelligent scene upgrade.

“Ai-based quality inspection products extend upstream, but also can backtrack the whole chain of production, improve the digital process of each link, to achieve the whole life cycle management of products. It will be a simple and complete digital factory.” “Huang Qiang said.

Since it serves as a gateway for diversion, Tencent’s expectation for AI quality inspection is that it will eventually become a universal product for different industries, which means that the cost should be reduced quickly. But Tencent was initially unprofitable in the first two cases. “We put 8-10 algorithm engineers on a project for more than half a year.” “Huang Qiang said.

According to the reporter of Southern Finance All Media, according to the market price of an algorithm engineer of 30,000 yuan/month, a project does not count the cost of hardware, server and so on, the bottom of 1.44 million yuan. However, the market price of a complete set of AI inspection equipment solutions is in the range of hundreds of thousands to small millions.

At this time, the cloud platform, which has been burning money for many years, has become the base for Tencent to reduce costs substantially. “The cloud is definitely a big advantage for us. The cloud is based on computing power, and when someone else runs on one [computing power] card, we run on 100, and the efficiency of the iterated model is definitely different.” “Huang Qiang said.

Under the support of the cloud platform, Tencent has been able to do, an algorithm to solve a project, the time has been shortened from half a year to 3 months. “Further down, when we only need 0.1-0.2 algorithm engineers to solve a project, the cost of upgrading AI quality inspection for enterprise customers will continue to fall,” Huang said.

In addition, Tencent also has a general product for small and medium-sized enterprises – AI quality inspection machine “Tenghui Fei Pupil”. In this general purpose product, Tencent will only be responsible for the software algorithm part, the rest of the hardware to other manufacturers. In this regard, Huang Qiang will be called Tencent Cloud “be integrated” strategy.

It is worth noting that although process industrial manufacturing scenarios can play the “industry general” card to reduce costs, there are limitations in discrete manufacturing. Huang Xiujin, CEO of Pinhui Intelligence, which focuses on AI quality inspection in the semiconductor field, said that AI quality inspection in the semiconductor field requires a deep understanding of the industry know-how. The reason is that the definition of semiconductor product quality is still in the post-standard period, and there are more than 20 criteria and hundreds of algorithms behind them in the midstream closed test process alone. At the same time, the accuracy of the optical imaging algorithm is also higher, and even the arc height of the chip wire needs to be detected by 3D means.

In terms of market size, only the AI quality inspection in the semiconductor field will be a market of 100 billion. “Take the semiconductor closed test step as an example, it involves at least four tests before and after. As the domestic semiconductor manufacturing market becomes more mature, the wave of domestic innovation will follow.

At present, Tencent also focuses on the product appearance detection stage based on the AI quality inspection case of general industry. Huang said Tencent is willing to cooperate with industry partners to make the industry pie bigger. In the niche areas that Dachang cannot reach, there is a window for startups that are deeply engaged in vertical industries to conquer.