Continuously adding downstream application companies, Nvidia is drawing a new investment map

Author: Zhang Xinyi Wang Xinhao

Source: China Electronics News

Original Title: "NVIDIA's Investment Map"

In the past two months, Nvidia has made frequent investments and acquisitions: first, it invested in three AI unicorn companies, then spent $50 million to support biotechnology, and acquired the AI cloud service company Lambda Labs on July 19. There is even more news that Nvidia is rethinking becoming an anchor investor for ARM after its acquisition of ARM failed. In the first half of this year, Nvidia, which has been continuously adding downstream application companies, is drawing a new investment landscape.

NVIDIA Investment "Three Steps"

According to public data, since the establishment of Nvidia in 1993, it has invested in and acquired more than 50 companies.

The development of Nvidia can be divided into three stages: the maintenance period from 1993 to 2006, the development period from 2007 to 2015, and the expansion period from 2016 to the present.

Data source: Compiled by China Electronics News

In the first stage, Nvidia, which focused on self-developed game graphics processors, changed the competitive landscape of graphics cards by annexing the number one competitor. During this time, Nvidia rang the bell, launched the GeForce256 and defined the graphics processing unit GPU. In 2000, Nvidia acquired 3DFX, a pioneer of 3D graphics cards, for US$70 million. Since then, after experiencing the "survival game" of graphics card companies in the 1990s, Nvidia has acquired its number one rival, 3DFX, and emerged as a survivor.

In the second stage, Nvidia's investment and mergers and acquisitions were not high-profile. In 2007, the establishment and application of the moat CUDA enabled Nvidia to build an ecological barrier different from other graphics card manufacturers. From the perspective of "hindsight", Nvidia developed the CUDA platform, explored general-purpose GPUs, and decisively gave up the mobile market. This series of actions profoundly affected the development direction of Nvidia, and finally promoted Nvidia to form a game display, data center, and automobile as a platform. The business route of the "troika".

Since 2015, Nvidia's investment and merger process has accelerated, and its support for start-up companies has also begun to increase, and it has entered a stage of rapid expansion. In the past 8 years, Nvidia has invested in more than 30 companies from many countries and regions, involving artificial intelligence, image processing, autonomous driving, biomedical and other fields.

Data source: Compiled by China Electronics News

Huang Renxun once mentioned Nvidia’s investment logic: first, the company’s vision is consistent with Nvidia’s, that is, to use AI technology to create more value for society; second, this company needs Nvidia’s help; third, this company has relatively excellent qualifications .

Another round of aggressive expansion

In 2017 alone, Nvidia invested in and acquired about 20 companies, which has exceeded the sum of the previous years. In the following years, Nvidia's road to mergers and acquisitions was not smooth sailing. It successfully acquired Mellanox, a provider of cloud network switches and adapters, but it also had twists and turns in the process of acquiring ARM from Softbank, and finally had to terminate the plan due to antitrust laws. Semiconductor expert Mo Dakang said that Nvidia's investment fields seem to be complicated, but they can be briefly summarized into two basic directions: one is artificial intelligence with GPU as the core of computing, and the other is automotive electronics.

It can be found in its investment list that the further improvement of GPU software and hardware capabilities and the exploration of autonomous driving have become the main tasks of Nvidia. There may be signs of such a long-term deployment much earlier. Sheng Linghai, an expert in the semiconductor industry, told the reporter of "China Electronics News": "Nvidia decided to expand the application scenarios of GPU, and GPGPU (General GPU) was born because Huang Renxun realized that in addition to games, the parallel computing that GPU is good at may also be used in other fields. In the future, it will be useful in other industries that need to process huge amounts of data."

At the end of 2022, the generative artificial intelligence headed by ChatGPT entered the public's field of vision. The huge demand for computing power drove the GPU market to be hot, and pushed Nvidia to achieve a trillion-dollar market value in 2023. This is not "getting rich overnight", but accumulating. The software and hardware ecology formed by the long-term layout has become the "confidence" of Nvidia. After an aggressive investment in 2017, Nvidia appears to be on the road to yet another expansion.

In the first half of 2023, Nvidia has successively invested in three companies known as "AI unicorns". Among them, Canadian AI company Cohere announced the completion of $270 million in Series C financing; another artificial intelligence startup company Inflection AI announced that it is developing a supercomputer equipped with 22,000 NVIDIA H100s. In addition, Nvidia's investment strategy also involves artistic and creative AI, and Runway said it will use artificial intelligence for video production. The AI sci-fi trailer "Trailer: Genesis", which has recently attracted attention on social media, was handed over to Runway for video generation.

Runway Gen-2, a product of Runway (Source: Runway official website)

On July 12, Nvidia said it would invest $50 million in Recursion Pharmaceuticals, an AI pharmaceutical company, to accelerate the training of its artificial intelligence model and use it in drug development. The two companies will cooperate to advance the development of Recursion's AI-based models in the fields of biology and chemistry, and distribute them preferentially to biotech companies using NVIDIA cloud services.

Extend to the vertical field

Nvidia is aiming at the layout of downstream applications this time, more to improve its existing AI software and hardware ecosystem, and to expand business channels while strengthening the connection between supply and demand.

Nvidia is familiar with the battle in the graphics processor market as a whole, and its products have been battle-hardened since the company was founded. The powerful computing power that the GPU can provide is the foundation of its existence, but it does not mean that continuous improvement of computing power will be safe forever. Behind the "AI fever" is even more "cold thinking". "AIGC is a topic full of unlimited imagination, and it has already shown its prominence in the fields of leisure and entertainment, shallow office. With the emergence of various special models and further improvement of computing power in the future, AIGC will penetrate into more professional technical fields." Deng Chuxiang, a researcher at CCID Consulting, told the reporter of China Electronics News.

On the one hand, in industries that process large amounts of data, CPUs are not capable of handling huge amounts of calculations, and GPUs are required to process a large amount of parallel data; The data is relatively closed and complex, and more dedicated models and dedicated computing power chips need to be developed for training and reasoning.

As a chip design company, Nvidia stands on the computing power supply side and needs to consider the issue of "who will sell the product to". This requires Nvidia to actively explore and actively meet the needs of downstream application companies. In terms of AIGC, autonomous driving and biomedicine, Nvidia has begun to continuously provide high computing power services to various manufacturers.

In addition, Nvidia needs to accelerate its own ecological penetration through multi-party cooperation. If the completion of the CUDA platform realizes the "use" of the GPU, then now Nvidia will promote this set of software and hardware ecology to the "application" level to achieve another extension of the business framework. Therefore, Nvidia not only provides equipment, computing power, and technical support to the downstream application industry, but is also an investor in downstream emerging companies, thereby achieving a spiral upward process of repeatedly strengthening itself.