Source: Barron's Author: Bill Alpert
People don't doubt the revolutionary nature of artificial intelligence, but the return on investment of artificial intelligence remains to be seen.
Investors all over the world are looking for it$NVIDIA (NVDA.US)$Artificial intelligence value — The stock price has been pushed above $1,000 since May 22. Data center builders spent five times more on Nvidia in the last quarter. In order to compete for what Nvidia CEO Hwang In-hoon called an “AI factory,” corporate capital investment has already been scheduled for next year.
What do these investors expect from it?
People don't doubt the revolutionary nature of artificial intelligence—just like the age of steam and electricity. The return on investment in artificial intelligence remains to be seen, though. Behind every stock analyst's target price, and every CFO's approved multi-billion dollar data center, there are expected returns.
Is Nvidia's return on investment really high?
At the earnings conference call held on May 22, Hwang In-hoon said that companies and even state agencies are using Nvidia's accelerated computing products (GPUs) to transform data centers worth 1 trillion US dollars.
According to Nvidia investors, for every dollar invested in Nvidia's computing and networking products, these data centers can earn $5 in four years — meaning an annual return of 50%.
According to Nvidia, the return on investment calculation is just an “illustrative example.” Due to the confidentiality of competing companies and the innovative nature of artificial intelligence models, it is difficult for people to obtain the actual return on investment in AI.
Currently, the market is based on OpenAI's ChatGPT or Alphabet$Alphabet-A (GOOGL.US)$Its Gemini products are rare;$Microsoft (MSFT.US)$Their Copilot virtual assistant costs $10-30 per month.
Furthermore,$Netflix (NFLX.US)$,$Amazon (AMZN.US)$And Google are using artificial intelligence internally to target content to search audiences, but these are operating costs rather than purchasing separately priced products.
One way to find the return on investment in artificial intelligence is through artificial intelligence itself. So we asked ChatGPT about the quantitative return on investment in artificial intelligence.
“Quantifying the return on investment (ROI) of artificial intelligence (ROI) can be complex and depends on a variety of factors, such as specific application scenarios, industry conditions, maturity of use, and organizational goals,” ChatGPT said. “However, in many cases, investing in artificial intelligence can be very rewarding.” ChatGPT added.
Chatbots under Anthropic and Perplexity gave similar answers to ChatGPT.
Google added an “Overview” summary to its search results list, generated by the chatbot Gemini. Gemini's answer is more rigid than the chatbot above: for AI programs, ROI is calculated by subtracting the cost of investment from the net return on investment, then dividing that number by the cost of investment and multiplying by 100.
The investment costs listed by chatbots include infrastructure costs for data centers, computers, and networks, and the cost of using artificial intelligence models with hundreds of billions of variables — data scientists need money, as well as data to train models to complete specific tasks. For example, Barron's owner News Group (NWSA) last week announced an agreement allowing OpenAI to use its news content.
The chatbot also lists the benefits (or rewards) of investing in artificial intelligence: cost savings, increased revenue and customer satisfaction, and reduced risk. Among them, cost savings ranked first.
This may help explain why it's hard to find real examples of AI rewards. ChatGPT said, “Artificial intelligence can automate repetitive or time-consuming tasks, saving labor costs. For example, artificial intelligence chatbots can handle customer queries, thereby reducing the need for human customer service.”
In other words, artificial intelligence could replace workers or affect their wages, and every chatbot has shown this. Organizations deploying artificial intelligence are unlikely to do this, though.
200 years after the Industrial Revolution replaced manual labor, America's unemployment rate is still low. Artificial intelligence may increase the productivity of knowledge workers without layoffs. If artificial intelligence replaced me, I wouldn't cry because it would create new jobs. Even so, the political impact of unemployment is less positive, making companies less willing to share the return on their investment in using artificial intelligence.
Back at Nvidia, it lists some jobs suitable for “getting a huge return on investment from AI”: 1 billion knowledge workers, 50 million content creators, 1 million legal professionals, 9 million educators, 15 million call center agents, and 1 billion agricultural and food workers. Let's hope these impacts are only in terms of benefits, not costs.
In addition to the innovative nature of artificial intelligence, their subsequent impact is harder to predict — new features of these models will appear as they scale, and may bring about things that were not anticipated at the beginning of the design.
For example, new medicines may appear; “wrong” governments may use artificial intelligence to cause harm; short-term unemployment of workers may also fuel extremist political tendencies.
At this stage, we can at least understand Wall Street's expected return on Nvidia stock. Analysts covered by FactSet gave Nvidia's average 12-month target price of $1,169, which is almost no premium over the stock price of $1148.25 at the close of May 29. Since 2022, Nvidia has generated a return on investment of over 800%.
edit/lambor