After attending GTC, Morgan Stanley said: We were wrong, we originally thought that physical AI "was still a story."

From concept to reality, physical AI investment is accelerating. After attending NVIDIA's GTC conference, Morgan Stanley has reassessed the investment prospects in the robotics and physical AI sectors.

In a report dated March 24, Morgan Stanley pointed out that after this year's GTC, analysts found that physical AI will have a substantial revenue impact on the semiconductor industry sooner than expected. Two key factors are changing this assessment:

• Investment is accelerating: Companies have begun to invest in developing models in the physical domain.
• Physical AI is becoming a core feature of the next generation of AI models: The next generation of AGI models under development will integrate physical simulation capabilities.

Previously, Morgan Stanley tended to believe that physical AI was still quite a long way from true commercialization, affecting stock valuation multiples rather than directly contributing to revenue. However, Morgan Stanley has observed that customer interest in robotics and physical AI has significantly increased compared to a year ago. Current physical AI is akin to generative AI from 4-5 years ago or autonomous driving from 7-8 years ago. In other words, companies are now starting to spend money developing models in the physical domain.

Key Conditions for the Development of Physical AI Are in Place, Semiconductor Companies Will Benefit

Morgan Stanley noted that companies are actively investing in the development of physical AI models.

These models require multimodal AI modeling capabilities that can handle visual, audio, and language data, and possess reasoning abilities. These capabilities have only begun to mature in the past few months. Companies have also emphasized the significant differences in investing in physical AI models compared to language or visual AI fields. They are actively investing in the collection of real-world data and creating simulated data, such as from NVIDIA's Isaac project.

As large model developers seek to further differentiate their models in the future, better integration of physical AI data has become a focus. Morgan Stanley expects,

Just as large language models (LLMs) have made progress in handling new types of data over the past 12 months, the next generation of AGI models will also incorporate physical simulation into their intelligence systems. Robotics startups are also leveraging large language models as a starting point for developing physical intelligence.

So who benefits, and by how much? Morgan Stanley pointed out,

In the field of physical AI, startups are raising funds on a scale of billions of dollars. Hardware will be the primary beneficiary of early investments, but considering that the market size for data center AI semiconductor processors will exceed $200 billion this year, smaller projects will not have a significant impact on the market. However, if we start to see a sustained increase in cluster scale, the situation may change.

Among them, NVIDIA is a core beneficiary, as most developers use NVIDIA for data simulation, training, and equipment building. Ecosystem participants, such as AMD, AVGO, and MRVL, will also benefit However, Morgan Stanley also expressed a cautious attitude towards "robots entering every household" in the short term, noting that the massive investments in data centers in the autonomous driving sector over the past few years have not yet fully translated into revenue. Morgan Stanley believes it is necessary to closely monitor the development of physical AI.

Risk Warning and Disclaimer

The market carries risks, and investment should be approached with caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this is at one's own risk