Is this a bubble, or the beginning of the next big thing? That's the question investors ask themselves when a new technology trend begins to take off.
That's definitely what's been happening in artificial intelligence (AI) stocks recently. On the back of this new technological capability, the tech-heavy Nasdaq-100 is up 39.5% on the year, and the iShares Semiconductor ETF is up a stunning 47%.
While some AI-related stocks may be a bit ahead of themselves, both of the aforementioned indexes are actually still below their all-time highs from late 2021. Meanwhile, new numbers and projections that have emerged over the past few week or so by industry participants indicate AI is likely the real deal, and set to create a lot of new revenue, profit, and economic value in the years ahead.
Image source: Getty Images.
Metric 1: 50%
While investors don't know exactly which large enterprise will be a biggest winner from AI, it appears the semiconductors that power AI are set for some big-time growth. These include leading-edge chip stocks, especially those that design and manufacture graphics processing unit (GPU) accelerators, which function as the picks and shovels to the AI gold rush.
Some recent growth projections on the AI accelerator market were given by Advanced Micro Devices (AMD 2.26%) CEO Lisa Su last week. At the company's AI chip presentation last Tuesday, Su estimated the AI accelerator market will grow at a 50% annualized rate over the next five years, from $30 billion in 2023 to $150 billion by 2027.
That kind of massive growth is why market leader Nvidia (NVDA -0.01%) has skyrocketed this year after forecasting blowout second-quarter guidance, and why AMD is also delivering strong stock performance despite relatively tepid near-term results.
In addition to Su's 50% annualized growth forecast, Nvidia just happened to forecast 50% sequential revenue growth for the current quarter, which beat analyst revenue estimates by, you guessed it, around 50%. Coincidence? Well, yes, that was probably a coincidence -- but no less impressive.
Metric 2: $1 trillion
Not only will the AI chip and server market grow, but Nvidia CEO Jensen Huang believes that most data centers will have to be reconfigured from a CPU-based traditional computing structure to an accelerated computing structure in the years ahead.
Huang pointed to about $1 trillion of traditional CPU-based data centers installed over the past four years, all of which he believes will eventually be replaced by accelerated computing systems. Assuming a four-year life for the average enterprise server, that's about $250 billion of data centers that will need to be replaced every year with updated AI models.
On the conference call with analysts, Huang noted:
Over the last four years, call it $1 trillion worth of infrastructure installed, and it's all completely based on CPUs and dumb NICs. It's basically unaccelerated. In the future, it's fairly clear now with this -- with generative AI becoming the primary workload of most of the world's data centers generating information ... and the fact that accelerated computing is so energy-efficient -- that the budget of a data center will shift very dramatically toward accelerated computing, and you're seeing that now. ... You're seeing the beginning of, call it, a 10-year transition to basically recycle or reclaim the world's data centers and build it out as accelerated computing.
Metric 3: $119 billion
All this investment in artificial intelligence will require more leading-edge logic and memory chips, which means more investment in chip fabs and semiconductor capital equipment.
2023 will actually see a decline in fab investment of about 18%, according to SEMI, a leading semiconductor industry association. However, off this year's trough of $74 billion, SEMI sees the semiconductor capital equipment industry catapulting higher over the next three years, growing 12% next year, accelerating 24% in 2025, then growing another 17.7% in 2026 to reach $118.8 billion that year.
Ajit Manocha, SEMI president and CEO, said in the press release containing the projection: "The projected equipment spending growth wave underscores the strong secular demand for semiconductors. The foundry and memory sectors will figure prominently in this expansion, pointing to demand for chips across a wide breadth of end markets and applications."
Remember, investment tends to lag the growth of the industry, as the downturn seen especially across PCs and smartphones last year caused the declines we are seeing in 2023. However, with AI demand now kicking into high gear, it looks as though AI will lead the chip industry out of its current slump, with this year's demand leading to a resumption of chip investments next year.
Image source: Getty Images.
Metric 4: $4.4 trillion
Finally, just last week, leading consulting firm McKinsey released a report on the effects of generative AI on the global economy. According to its updated predictions, McKinsey now believes generative AI will create up to $4.4 trillion of new economic value, mainly due to the effects from increased automation and worker productivity.
For context, the global economy was roughly $104 trillion last year; therefore, generative AI could add about 4% to the world's economy just by itself, which would be a significant boost.
According to the report released on June 14, generative AI could help boost knowledge worker productivity by 60% to 70%, with half of all work becoming automated between 2030 and 2060. Of the knowledge worker productivity and automation benefits, most will come from automating a business's operations and customer support through chatbots. Meanwhile, software engineers will benefit from automated coding, and research and development organizations will see a jolt of productivity through fast insights from enormous historical data sets.
Of course, a lot of these trends have already been underway for some time, so they shouldn't be a surprise; however, the McKinsey report is unique in that it attempts to quantify the effects of generative AI in terms of dollars and cents.
More likely than not, however, this prediction won't be entirely accurate. The update is a big leap forward from McKinsey's prior projections, so it's possible the effects of AI could will be even greater as technology develops. On the other hand, any negative use of AI, such as for misinformation or cybercrime, could potentially also harm the global economy on the other side of the equation.
Still, it appears as though AI is going to create a lot of economic value in the years ahead. While the current run in AI stocks may seem like another overenthusiastic speculation to some, there appears to be a real economic tidal wave coming through the rest of this decade.
