The hype train for Artificial Intelligence is full steam ahead but financial experts and researchers are disagreeing about its financial returns. AI is being touted as the current golden child in the tech sector where it has been the financial boon for NVIDIA and its CEO Jensen Huang causing virtually every other tech manufacturer to jump into the pond, along with their investors but some are becoming concerned that that financial returns may not be all they’re cracked up to be.
A report from Goldman Sachs states that over $1 trillion is expected to be spent on AI in the following years but then venture capital firm Sequoia Capital, per Tom’s Hardware, has said the industry will need $600 billion annually just to break even with its investments. A massive stock sell-off occurred this week on July 11 that was tagged the event as a ‘wake-up call’ as financial experts warn of a potential impending bubble burst for the AI industry.
Per Investor’s Business Daily:
“Yesterday was the wake-up call many expected and wanted in order to start taking at least some profits in Mag 7 tech and semi AI winners,” Mizuho Securities trading-desk analyst Jordan Klein said in a client note Friday.
“Yesterday’s semi sell-off (was) a quick preview of what is to come when Nvidia finally guides only ‘in line’ or misses the whisper (number) by a real amount,” Klein said. “Yes, that day will eventually happen.”
NVIDIA and other stocks did, somewhat, bounce back on Friday as is typical of corrections during changes in the market but the event was well noted. The report from Goldman Sachs exposed division among its own teams as some believe the AI industry is going through expected growth patterns while others are doubtful that financial returns for AI are limited due to it being unable to solve meaningful complex problems more efficiently than present technology. Estimates for output from AI in the report are at 9% productivity increase with a GDP increase of 6.1% but MIT Professor Daron Acemoglu sharply disagrees with an estimate of 0.5% and 1%, respectively.
Per Goldman Sachs:
“The forecast differences seem to revolve more around the timing of AI’s economic impacts than the
ultimate promise of the technology. Generative AI has the potential to fundamentally change the process of scientific discovery, research and development, innovation, new product and material testing, etc. as well as create new products and platforms. But given the focus and architecture of generative AI technology today, these truly transformative changes won’t happen quickly and few—if any—will likely occur within the next 10 years.”
Jim Covello, Head of Global Equity Research at Goldman Sachs, also shared concerns with the expected costs to develop AI. Covello explains that the gains will depend on how AI is used.
“My main concern is that the substantial cost to develop and run AI technology means that AI applications must solve extremely complex and important problems for enterprises to earn an appropriate return on investment (ROI). We estimate that the AI infrastructure buildout will cost over $1tn in the next several years alone, which includes spending on data centers, utilities, and applications. So, the crucial question is: What $1tn problem will AI solve? Replacing low wage jobs with tremendously costly technology is basically the polar opposite of the prior technology transitions I’ve witnessed in my thirty years of closely following the tech industry.”
The future looks so bright
However, some experts believe that AI is moving forward at a good pace and is even gaining traction as its technology progresses. This could be thought of as a water-is-wet analogy but technology doesn’t always manage to deliver on promises made by its engineers, designers, or investors. Time, supply chains, manufacturing processes, and more, can play crucial factors in getting things off the ground and running. Kash Rangan, Senior Equity Research Analyst at Goldman Sachs shares optimism for current investments and notes the differences with it compared to previous missteps made in the 1990s.
“Leading the late-1990s investment cycle, by contrast, were companies that didn’t have the financing, reputation, or knowledge to succeed, resulting in a tremendous amount of underutilized capacity. The companies spearheading the current investment cycle are also run by very capable managements, with CFOs watching expenses like a hawk, holding companies accountable for the return on investment, and standing ready to tap the brakes on spending if the returns disappoint. Of course, this could all still fail, resulting in the loss of tens of billions of dollars in capex and interest income. But the opportunity costs of pursuing these strategies despite unknown outcomes still seems small compared to the potential opportunity of building the foundation for the next big computing architecture”
Discussion (7 replies)
Join Discussion →Gee when you need to spend 3 million + on a custom water cooled rack that ROI gets difficult to prove. ESPECIALLY if you have to refresh the hardware every 3 years just to keep the hell up.
We will see a crash in the next 3-4 years after MS proves they are losing money for their copilot program. We are in the rising curve, so make the most money you can now and jump off when it starts to dip.
Yeah, AI isn't going anywhere. It just needs to come down to earth a little and start getting realistic expectations.
By the sounds of things, that is starting to happen.
During the dotcom bubble lots of companies had webpages because "the information super highway is the future. www yallz!" But they weren't actually doing anything useful with them. In many cases you couldn't even find a brick and mortar stores hours on their webpage. It was more like some kids first webpage. "This is my webpage, let me tell you about myself". But no one cared.
I feel like that is where wee are with AI right now. Sure, some companies are doing interesting and value added work with AI, but then there are LOTS that are just doing AI because AI is trendy, and they don't even know why they have AI. That needs to change, and it likely will soon.
The sad thing is Ai will catch on for businesses as soon as they reliably and successfully replace humans with it. When an AI rack can replace 800 call takers... then we're talking about an impact on business.
And a impact on those 800 people now without a job, insurance, stability.
It's going to be a freaking MESS when this hits hard.
While I am not a huge fan of AI (I find it annoying, and with the potential to cause all sorts of problems if we trust it) I am also not in the doom and gloom camp.
Every single time we as humans have advanced the technology in the labor market, some people have been hurt short term, but then the labor market has adjusted, vastly increased in productivity, and the economy and wages have grown as a result.
I think the same will happen with AI. I doubt it will ever be able to replace humans completely, but it will likely get to the point where it - just like when we first got computers on our desks - is a technology that aids human employees and makes them more productive.
This will result in fewer people being needed for the same work, but on the flipside this will also lower the cost of that work, allowing the product or service that company sells to be cheaper, expanding its market, etc. etc, growing markets and the economy and creating other jobs for those who lost them.
This happened with an industry I used to work in. (though I worked there long after this happened).
I used to work in the design and development of HPLC systems. Back in the 80's chromatography was hard. It took a Chemist with an advanced degree packing his own columns and waiting hours or even days for a single sample to gravity feed through and be analyzed by some sensor on the other side.
Then they found a way to do it at increasingly high pressures with pre-manufactured columns and increasingly effective sensors, and auto-samplers, etc, etc. all controlled by a computer. A single desktop could over a network control a room full of HPLC or UHPLC systems. What used to take a PhD Chemist a week or a month, could eventually be condensed down to a couple of minutes per sample, and auto-samplers enabled these systems to run 24 hours a day. And you really only needed one Phd Chemist to configure a particular test, commit it to code, after which tech-level analysts could do much of the work just executing the program and configuration the Chemist had created.
Did this result in mass unemployment in highly skilled Chemists? Nope. Instead a single well paid chemist could (with an equipment budget and some help from techs) output hundreds of thousands of analyses a week. This drove down the cost of chemical analysis making it affordable in places it had never been before. Demand exploded due to the lower cost. More and more analytical chemistry labs started popping up. We were testing everything from water to food to crime scenes to medical tests that previously had been cost prohibitive. It even allowed a whole new industries in Biotech and Pharma to blossom in ways they previously couldn't have without the access to affordable and quick analytical QC.
Demand for highly skilled chemists surged. Their work was a little different now than it had been in the past. The fact that they were not manually packing columns and monitoring sensitive equipment allowed them to focus more time on where they truly added value, the brain work. Demand for analytical QC technicians to maintain and run the equipment also surged. Demand for the equipment surged, resulting in a surge in demand for everyone involved in manufacturing of the equipment as well.
Now, this is definitely a "best case" scenario for what AI will be able to do for us. In this case there were very few downsides. Very few people actually lost their jobs. Their jobs just changed, and they were in even greater demand over time.
With AI there will likely be pain as the market adjusts. Many people will likely lose their jobs and find that they no longer have the requisite skills to fill the positions that are now required to make a good living in the new market. And that sucks. We need to be prepared for that, and be much better at retraining people for future employment and keeping people from losing their homes than we have been traditionally or we are going to have HUGE societal problems.
Once we come through on the other side though, the economy will likely be even stronger than it is today with workers doing even better than they are today. This has been the case every single time technological advancement has upended any job market. The industrial revolution, computers, automated technology, etc. etc. These things were all difficult when they were new, and caused lots of people harm, but once we got through to the other side they changed our society in ways we could otherwise never have had, raising our standard of living by making products and services that were previously impossible (or at least impossible for normal people who weren't wealthy) possible for almost everyone.
I wouldn't want to go back to a pre-computer world, or a pre-industrial revolution world. And I'm sure in 50 years time the people living in an AI normitive world will feel the same way, and not want to go back to the way things were before.
Getting there will be difficult through, and I worry a little about this.
I also worry that we will wind up trusting AI too much in the interim, allowing it to make bad decisions for us that cause engineering disasters or financial crises, before we learn the hard wauy the types of controls and reviews we will need to have on the technology. (because - as a species - we always seem to need to learn the hard way)
The most difficult adjustment - at least in the U.S. will likely be the political one. In a market where marketable skills are likely to change very fast, and more and more of the ability to sustain a living will come from the ownership of the AI systems we may be forced into seriously rethinking our free market models. We are going to need vastly expanded social safety nets both to help people survive without losing their homes when their skillsets inevitably become obsolete AND to train and prepare them for the next needed skill that industry needs. If we don't, we will likely head for a combination of poverty and malaise, and industries screaming for talent they can't find.
This new society behooves everyone to drop their old political left/right biases and be open to a new way of equitably sharing the proceeds of the productivity of mankind. I don't claim to have the right answer of how that should work, but I do know that if we do nothing we will fail. And I don't have any confidence at all that people will be able to overcome these traditional political biases in ordser to adapt our society to the mostly inevitable future.
Some countries will adapt well. Those that already show a history of taking care of their citizens will implement a supported universal minimum income method of some sort. Others like the US will have a greater wage/class divide where those that manage to get support will get the bare minimum. At least that's how I see it.
I'm not convinced universal minimum income can ever work. At least as long as certain necessities (such as housing) have pricing set by supply and demand and are supply limited.
If you were to give ever last American today a $5k per month free basic universal income most would just use it to outbid each other on housing.
Property values would go up, rents would go up, and most people would be just as poor as they were before you gave them $5k per month.
There would have to be some pretty serious structural changes to the economy to make something like that work. You could try to control housing prices, but pretty much all that does is result in shortages. Seriously, every single country/city I've been to where there is rent control people are always complaining how difficult it is to find an apartment, and how long the wait lists are.
We have some real structural societal problems to solve. There is no silver bullet. Universal Basic Income definitely is not it.