There were more announced layoffs in the U.S. year-to-date through September 30, 2025, than in any other full year since the government shut down the economy in 2020, according to Challenger, Gray, & Christmas. To make matters worse, U.S.-based employers announced even more cuts in October 2025. Those cuts hit the highest level for the month since 2009, when the economy was reeling from the Financial Crisis.
In 2025, thousands of companies announced nearly one million layoffs in technology, corporate, and other roles, often citing artificial intelligence (AI) as a factor. That list includes companies such as UPS, Salesforce, Microsoft, Intel, Blue Origin, and Amazon. (Although Amazon has tried to walk back their AI layoff comments.) Other companies, such as Walmart, Ford, J.P. Morgan Chase, and Amazon, have been forthright in stating that although they are not yet able to implement AI widely, they do expect AI to enable them to eliminate jobs. Layoffs happen every year—that is nothing new. However, citing AI as a reason for layoffs is a relatively new development.
In many cases, AI is taking the jobs of young, entry-level workers. AI has also dented the demand for more experienced white-collar employees (back-office administration, lower-tier legal tasks, and basic software development); this is not going to stop anytime soon.
Yet, the economy has been humming along nicely. Even with missing important government data, I see that the Gross Domestic Product (GDP) growth rate is annualizing at roughly two percent. That is not gangbusters, but given the poor state of the labor market, one might expect flat to negative growth. So, what is keeping the economy afloat even though people are finding it more difficult to find jobs?
I recently attended another conference on how artificial intelligence directly intersects with my industry—money management. There were some tools for investment selection, but none of them were particularly effective, to be frank. They were good, but no better than all the stuff I already have access to. But I will circle back to that idea—AI for investment selection help—in a minute.
Most of the software tools at this financial conference focused on efficiency. For example, a widely implemented tool is note-taking devices that enable financial advisors to put down the pen and remain fully present during a client conversation. The software does not need to record the conversation—who has the time to go back and listen to an hour-long discussion, even if played at 1.5 times the speed? However, it can summarize the conversation more effectively than a human would, create an action plan, transfer those notes to the contact field in our client relationship management (CRM) database, and then easily recall those pertinent notes when reviewing financial management decisions or preparing for future talks. Additionally, there are other technologies available to assist with paperwork, compliance, and even highly sophisticated tax strategies—each of which helps remove friction and save time.
Back to the investment side, the output is no different from what I have used, but the time savings are valuable. So, even when the software does not directly improve the impact on the client, the secondary effect—having more time to spend with the client—does. So, with the implementation of AI, I could a) fire advisors, b) help more clients, c) go deeper with current clients, or (my favorite) d) go deeper with more clients without having to add to employee headcount.
Mine is not the only business seeing AI opportunities unfold, which is why the labor market has been weak this year, and it will likely remain so. Approximately 12,000 baby boomers are retiring every day, and companies are in increasingly less need of hiring replacements because AI tools can step in and perform the job or free up the time of other employees to take on additional work. And just wait until humanoid robots become mainstream!
Take a look at the videos from Boston Dynamics, Tesla, and Unitree to see how advanced humanoid robots are becoming in terms of ambulation and fine motor skills. Humanoid robots utilize large language models (LLMs) as their brains and train in a manner similar to ChatGPT, Claude, and Grok—by observing human behavior. The robots will be around humans, record and learn from them, and become brilliant.
However, having a robot with the sentience of Rosey from “The Jetsons” will take a considerable amount of time. I mean, you know we have self-driving cars now, right? But their use is not widespread. If you want to take a self-driving Waymo taxi, you are only going to be able to do that if you are in L.A., Phoenix, San Francisco, Austin, or Atlanta.
However, the first autonomous coast-to-coast self-driving car journey was undertaken in 1995 by Navlab, a project of Carnegie Mellon. Technically, about 98.2 percent of Navlab’s journey was autonomous. It was not until 2015 that Delphi Automotive undertook a similar trip, covering 99 percent of the distance. That proverbial (and in this instance, literal) last mile is the slowest.
You can buy robots for your house today for about $30,0000, but they still require lots of training from you. In a handful of years, you will see more of them in warehouses and probably on the battlefields. Use your imagination after that—mine has high-earning households adopting humanoid robots more broadly than Alexa or Google Home by 2032 (and that is being conservative).
Whether it is AI on your computer or AI in robots, I do not foresee the sort of AI-jobspocalypse clickbait headlines have predicted. With every industrial revolution, new jobs are created that we never thought of. Did you know that 60 percent of jobs today did not even exist as jobs in 1940? Since 1980, there has been a significant shift in new-job creation away from middle-pay production and clerical roles toward mid-pay service jobs and high-pay professional jobs.
How is it that innovations foster new jobs despite displacing others? Lower costs and new capabilities expand markets, spawn industry complements, and raise incomes. Together, this creates new tasks (jobs are really just bundles of tasks) and new industries even faster than old ones shrink—at least over time.
New technologies make goods cheaper, leading to increased purchases of these goods and their related products. For example, after the spinning jenny was invented (a multi-spindle machine for spinning wool and cotton), textiles were more readily available. The assembly line led to the mass production of cars, and the personal computer led to widespread internet use, ultimately giving rise to cloud software.
Then there are complementary jobs that support new technologies, including installation, integration, cybersecurity, user experience, compliance, training, maintenance, and sales. Electrification eliminated many steam jobs but created new opportunities for electricians, appliance manufacturers, grid operators, and inspectors. And you can thank the invention of the automobile for road construction, motels, and auto insurance.
While it will take some time for new jobs to catch up with the lost ones, the good news is that the overall economy should hold up OK until then (absent unrelated shocks, of course). The extra productivity from AI will help the economy. An approximation of GDP growth is the sum of the growth of total hours worked and productivity.
Real (inflation-adjusted) GDP growth
=
Total hours worked growth + Labor-productivity growth
The “prime-age” workforce (ages 25 to 54) growth is a good proxy for total hours worked (yes, that would be a heuristic within a rule of thumb). Note that this calculation is not intended to be exact.
Total hours worked have grown one percent year over year, according to the Bureau of Labor Statistics. And labor productivity has increased by 1.5 percent year over year. Together, that would suggest a GDP growth rate of 2.5 percent (it was approximately 2.3 percent, but we will not know for certain for a while as the government shutdown has delayed some official data).
Suppose the rate of hours worked is reduced from 1.0 percent to 0.6 percent, the lowest rate of any year in the last decade, except for 2020, when the government shut down the economy.
And let’s say the productivity growth rate increases from 1.5 percent and rivals the average rate of the 1990s. In that case, the rate rises to 2.12 percent (which is conservative, given that the productivity growth rate over the last three years, after workers more broadly adopted the productivity tools of that era, was 3.1 percent).
The sum of these two (0.6 percent from hours and 2.12 percent from productivity) would suggest a 2.72 percent GDP growth rate. That would be pretty good. While there will be a lot of unfortunate disruption to the workforce, especially for younger, entry-level employees, I see enough potential on the other side of the interruption to stay in the stock market. That being said, the S&P 500 has not had a five percent pullback since the April 2025 tariff tantrum; an overdue short-term correction can occur simultaneously to intermediate-term fundamentals playing out.
Allen Harris is an owner of Berkshire Money Management in Great Barrington and Dalton, managing more than $1 billion of investments. Unless specifically identified as original research or data gathering, some or all of the data cited is attributable to third-party sources. Unless stated otherwise, any mention of specific securities or investments is for illustrative purposes only. Advisor’s clients may or may not hold the securities discussed in their portfolios. Advisor makes no representation that any of the securities discussed have been or will be profitable. Full disclosures here. Direct inquiries to Allen at AHarris@BerkshireMM.com.
