Artificial intelligence will replace scientists, and deep learning will create robots that read. It’s crazy, true and coming faster than most investors believe.
That lede was created by GPT-3, a generative pre-trained transformer AI that editors at the MIT Technology Review believe will be among the most disruptive technologies of 2021. It’s scary good.
The idea of computers that learn, read and think has been around for years. The biggest roadblocks are the limits of computer processing power and the lack of labelled data AI algorithms need to learn from. The first hurdle is being overcome by advances in cloud computing, and the second is being tackled by GPT-3. It was designed from the ground up to autonomously label the billions of digital documents generated for use on the web.
GPT-3 comes from Open AI, a project founded by Elon Musk and Sam Altman, the founder of Y-Combinator, the developmental lab behind Airbnb, Inc. (Nasdaq: ABNB), Reddit and hundreds of startup companies. Together, the pair invested $1 billion to ensure that AI was being used for the betterment of mankind. The growth of the project has been exponential.
Related Post: Believe in the Digital Transformation
According to the original white paper, the first GPT crawled 7,000 books and generated only 18.4 machine learning tokens, the inputs used to advance algorithmic learning. GPT-3 slinks across the entire web and has collected 175 billion tokens so far.
The capabilities of GPT-3 currently range from translation to creative writing. And while that may seem unimportant, consider that AI can translate both spoken languages and code. This means GPT-3 can autonomously write actual code based on simple language input. Add creative writing … and the potential is unlimited.
In theory, these AIs can read, postulate and even write brand new code to execute those ideas. It’s exciting and scary at the same time because this is not general AI. The algorithms can put ideas into context, but they are not thinking, at least not in the human sense.
MIT Review editors are also quick to point out that GPT-3 is not a panacea. It is crawling the internet for inputs, a vessel filled misinformation and dangerous biases. This peril has not been lost on product managers at Open AI.
When the scope of the GPT breakthrough became apparent in 2019, the research was initially withheld from the public for fear it might be misused, according to a report from the Verge.
The investment angle is more straightforward.
Too often, investors fantasize about smaller businesses. They buy into the hype that tiny companies will disrupt slow-footed behemoths, creating the next Amazon.com, Inc. (Nasdaq: AMZN) or Alphabet, Inc. (Nasdaq: GOOGL).
In reality, AI is still the early stages, but NVIDIA Corp. (Nasdaq: NVDA) is best positioned to capitalize on this trend.
NVIDIA makes the powerful deep-learning infrastructure that is becoming foundational to the development of AI. The Santa Clara, Calif.-based company has been slowly transforming from its base in computer gaming to a vertically integrated AI platform with cutting edge silicon, software and the firmware and hardware bits and pieces in between.
During Q4, the company had $5 billion in sales, the first in its 22-year history. The 61% year-over-year growth in the quarter mirrors spectacular full year sales of $16.7 billion, up 53%. Gross margins were 65.5%. And profits rose 53% to almost $1.5 billion.
Related Post: Buy … Don’t Fear Tech Weakness
All of this was planned. Under CEO Jensen Huang, NVIDIA has been forging ahead with a larger plan to bring the company’s expertise in graphics to hyperscale data centers. Today, its CUDA software, Mellanox switches, cables and adaptors and continually evolving line up of high-performance graphics processing units are commonplace in cloud computing.
During last quarter, NVIDIA announced collaborations with Google Cloud and Amazon Web Services. These arrangements contributed to $1.9 billion in data center sales during Q4, up 97% from a year ago.
Moving more NVIDIA GPUs and other company gear to the cloud accelerates the development of AI. It means more powerful processing, and eventually lower compute costs.
Investors should consider taking advantage of the current NVIDIA share price weakness. It most likely will not persist.
Best wishes,
Jon D. Markman
