As an example of the innovation in artificial intelligence going on now, Google is working on technology to classify street signs, identify objects in photos and translate speech between two languages in real time. The company, in a product event showing off these technologies recently, admitted that this is just the beginning.
Stephen Chin, an analyst at UBS, agrees with Google and sees huge potential growth in the sector, especially in the silicon that makes it possible.
"The democratization of AI now underway could drive a new phase of growth in Semis that our new proprietary model estimates could represent a $35B revenue total addressable market by 2021 (ex-memory), or a 41% compound annual growth rate from current levels," Chin said.
The AI that currently exists is relatively rudimentary compared to its future potential, but even so, it requires huge amounts of computing power. Computer chip makers have realized this, and are currently in a race to provide the chips that can power current and future AI systems.
Graphics processing units, like those made by AMD and Nvidia, will make up 54% of the total AI market in 2021, according to Chin, while central processing units, like those from Intel and AMD, will make up about 40% of the markets. Chin said that data centers will make up a majority of the demand for AI-focused chips in the future, as smaller consumer computers won't have the power to train AI systems in a reasonable time frame.
With that in mind, Chin says that Nvidia is the company to beat in the space.
"We estimate Nvidia has a time to market advantage of 1+ year and should be able to sustain this lead," Chin said.
Nvidia has been on a world tour in recent weeks, announcing new AI-themed partnerships with drone makers, car companies and more. The company was crowned the smartest company in the world by MIT this year, in part, because of its dominance in the AI space.
Nvidia is also focusing on providing the tech necessary to power autonomous cars, another AI growth area, according to Chin. The company's "Drive" series of chips are already being used by a large number of car manufacturers.
Tesla, one of the biggest names in self-driving cars, is reportedly working with Nvidia's competitor AMD to produce a custom self-driving car chip, though. The autonomous car space could be a huge market for chip makers, coming in just behind data centers, said Chin.
Nvidia and Intel are the only two companies Chin sees as having a solution for every part of the current AI market. Most of the other companies, like AMD, Qualcomm, Samsung and even Google, fail to capture the entire breadth of that market.
This is an important advantage for the two companies, though it's not everything.
The computing ability of these chips is important, as the faster a chip is, the faster it can crunch through the huge data sets required to create artificial intelligence systems. AMD is working on CPUs that can rival Intel's and GPUs that can compete with Nvidia's. Chin said it could start to gain market share because of this.
In 2017, Chin estimates the total market for machine learning and artificial intelligence to be about $8.2 billion. By 2021, Chin thinks it could be worth upwards of $35 billion. The companies that can provide the best chips for that growing market stand to benefit from the sector's explosive growth.
Here's how much each of the chip makers is up this year, and their share prices as of about 3 p.m. ET Tuesday:
- Intel (INTC): +8.58%, $39.75
- AMD (AMD): +25.60%, $14.20
- Nvidia (NVDA): +93.74%, $197.51
- Google (GOOGL): +25.09%, $1,009.19