Loading summary
Transcript42 lines
- [00:00]
Bel Lynn
Amazon Business is 10 years old.
- [00:02]
Amazon Business Executive
We needed to build something that could address the unique purchasing challenges of different organizations. The need has never been larger and.
- [00:12]
Greater to use AI.
- [00:13]
Trying to manage that and reduce complexity. It is a fundamental sea change that is happening.
- [00:19]
We wanted to manage the number of suppliers. We were able to timely and accurately get the supplies to the end user's possession. Amazon solved that problem for us.
- [00:29]
Stephen Rosenbush
Foreign.
- [00:34]
Danny Lewis
Welcome to Tech News briefing. It's Tuesday, December 30th. I'm Danny Lewis for the Wall Street Journal. This week we're taking a deep dive into the future, exploring how the biggest shifts in technology will play out in 2026. As we look toward the new year, one thing is certain. The AI revolution is showing no signs of slowing down. Over two thirds of CEOs plan to spend even more on AI next year. That's according to an annual survey of more than 350 public company CEOs from the firm Teneo. That doubling down comes amid a rash of investor skepticism that companies are overspending on AI, sparking fears of a bubble and market turbulence as the year was winding down. So where does AI go from here? What are the next major battlegrounds for regulation, data and energy demands? And how will it all impact jobs? We have WSJ reporters Bel Lynn, Chip Cutter and Enterprise Technology Bureau Chief at the WSJ Leadership Institute, Steven Rosenbush with us to discuss. I want to begin by looking back at this year in AI. What were the standout moments for each of you? Bel, let's start with you.
- [01:45]
Bel Lynn
Sure. I'll start with the most recent event that comes to mind, which was just earlier this month in December, where OpenAI released this code red memo. And it was this really profound moment to me because it indicated that OpenAI, which was by far the sort of undisputed leader up until kind of recent times in its frontier AI models was really threatened by Google's Gemini models. And then going backwards in time a little bit, I also think about this fall where there was this flurry of what's called circular AI deals between companies like, like OpenAI, Nvidia, AMD, Oracle, Microsoft. And to me, that really underscored the risk of an AI bubble and really showed that there's a lot of big money at play and it's all interconnected. And then in the summer, in August, There was this MIT study which found that 95% of AI pilots at companies fail. And so there's a little bit of debate about over the sort of findings of the study, but it really highlighted for companies that AI isn't really showing its investment so far. And the fact that this technology could be overhyped, and that added to the fears of the AI bubble.
- [03:00]
Danny Lewis
Steven, what jumps to mind for you?
- [03:02]
Stephen Rosenbush
I'd like to highlight a few things happening beneath the surface, or at least out of the spotlight. If you look deep into the infrastructure of AI, I'm sure everyone wants to go there. You see that Google is now processing more than 50 times the amount of AI measured in these units called tokens, than it was last year. That's just an enormous increase. Think about how quickly or slowly the economy is increasing. 50 times is a big number. It doesn't appear to be slowing down. It appears to continue accelerating. And then Disney, it's licensing its characters for use on OpenAI. It follows other big companies, including Walmart, onto that platform and their efforts to create the infrastructure for commerce and payments on that OpenAI system. So when I look at that, I think the development of that economy, the development of that ecosystem is going to be really powerful and that over the long term, all these other things, which are very real, are going to be driven by the creation of that economy.
- [04:10]
Danny Lewis
And chip, you've taken a look at how AI is already impacting jobs. Talk to me a little bit about what that looked like in 2025.
- [04:18]
Chip Cutter
This was the big question all year. It's like, when are we going to start to see AI lead to big job cuts? Because for a long time you had chief executives making this case that AI was just going to change the roles or that they would just be able to do more with their existing teams. And we started to hear the language and the action shift. A notable moment for me came in the middle of this year in June. And that's when Amazon CEO Andy Jassy wrote a note to the company saying that as we roll out more generative AI and agents going to change the way our work is done. He said that it's hard to know exactly where this nets out over time, but in the next few years, we expect this will reduce our total corporate workforce. For a lot of employers, that was a huge moment. Amazon is, of course, one of the biggest employers in the us. If you have the leader of one of the most prominent companies saying, wait, we think our corporate workforce can be smaller. A lot of other executives are going to take their cues from that. We've seen that. We're seeing more and more companies now be open about the fact that they will not need to hire as many people. They will be able to let people Go. And we're starting to obviously see this ripple of layoffs start to happen among some of the most prominent companies in America.
- [05:20]
Danny Lewis
What does 2026 have in store there then?
- [05:23]
Chip Cutter
We're already starting to hear from companies saying that they can keep headcount flat. But I think what we're also going to see is a more willingness from companies to let people go. CEOs were often pretty reluctant to say they were going to do layoffs. Now we've seen companies say that they almost feel like staffers are an impediment to growth, that having a bigger headcount actually hurts a company's ability to reach its goals. It's a real shift from, from just a few years ago when companies used to hoard talent, right? Coming out of the pandemic, they would hire people even without work for them to do. And now we're at this moment where companies just want to get as small as possible. CEOs look at some of these startups that have a dozen employees and billion dollar valuations. They look at that with envy and say why can't we do that? Why can't our workforce just be a lot smaller aided by these AI tools?
- [06:05]
Danny Lewis
So what about all these emerging new roles that we were told AI would be creating?
- [06:09]
Chip Cutter
We're certainly seeing some of that already. Walmart is an interesting case study here actually because it's the country's largest private employer, big retailer, not a technology company. The company has said that it sees AI changing literally every job. That means it's going to eliminate some roles but also create others. So Walmart, for example, created an agent builder position this year. That's an employee who helps build AI tools to help merchants. It also said that it expects to add people in sort of high touch customer areas like its bakeries, but at the same time it's been able to automate some back of store tasks with the help of AI and expects more changes to come. Like a lot of companies, Walmart brought on an executive who reports to its CEO who is going to think about its AI ambitions and how the company changes.
- [06:52]
Danny Lewis
Steven, what about the impact of AI agents?
- [06:55]
Stephen Rosenbush
This was in theory supposed to be the year of the AI agent. It wasn't quite that. We've seen development of AI agents both in terms of infrastructure. There was some important technology gains protocols that came out of Google and Anthropic that really helped build all of the tooling and infrastructure around the models at the heart of these agents, which are really complicated systems. We're not there yet. I certainly don't think that 2025 will go down in history as the year of the agent. I'm not sure that 2026 will either. But when I talk to big companies, they are using agents. They have not gotten to the point where they're ubiquitous and they're like transforming the company on a wholesale level. But adoption is growing. I think we'll see more relatively significant, but not overwhelming incremental growth in the coming year.
- [07:52]
Danny Lewis
What's your take on the changes we've seen in AI this year? If you're a listener on Spotify, let us know in the comments. Up next, we take a look at what's in store for 2026, from the companies at the center of the technology to the energy demands ahead. Plus, is there a glimmer of hope for new jobs created by AI? That's after the break.
- [08:17]
Amazon Business Executive
We wanted to manage the number of suppliers, get better visibility and transparency into these small transactions.
- [08:24]
We realized that businesses and organizations of all sizes really wanted the same things that consumers did. They wanted a massive selection, great prices and great value. They wanted that delivery experience that they get from Amazon in their personal lives. So they really wanted the Amazon experience but for work.
- [08:53]
Danny Lewis
And we're back with WSJ reporters Bell, Lynn and Chip Cutter and bureau chief Stephen Rosenbush to continue our conversation about the future of AI. This was a year where we arguably saw dominant players emerge in the race for AI supremacy, and their positions shifted a bit as the year went on. Bell, walk me through what happened and where we stand heading into 2026.
- [09:15]
Bel Lynn
Sure. So I'll start with OpenAI, which as I mentioned at the top of our conversation, really led the market with its Frontier AI models. But it's gone through a sort of turbulent 2025. And so overall what that means is that its status as by far the market leader is no longer a guarantee. A lot of companies might say a few years ago that they would reach for OpenAI models without question, but we're increasingly seeing companies use a bunch of different models from Anthropic as well, from Cohere even. And so what we're really seeing is almost a much more open market for a new leader in a frontier model.
- [09:53]
Stephen Rosenbush
I would say that yes, we've seen increasing competition over the last year, especially in the area of model development, also in the area of chip chips, GPUs, where Google in particular is putting some pressure on Nvidia. Overall, that's a good thing. It should drive down the cost of computing of creating those tokens, and that in Theory should drive up consumption. There's this idea in tech of this thing called Jevons Paradox, which means that as the price goes down, people don't just say, oh, that's great, I only need one box of cereal. I'm happy to just pay less for it. They'll buy two boxes of cereal. So in theory, it should increase consumption. We'll see.
- [10:37]
Danny Lewis
Data and energy demands are also huge issues when it comes to the future of AI. So what should we expect 2026 to bring in those fields?
- [10:45]
Stephen Rosenbush
The energy demands for AI are somewhere between ginormous and astronomical. It's very, very difficult to meet all of that demand without throwing everything at it. So I think that we'll see more nuclear power. Three Mile island plant is being revived by Microsoft. There's a lot of innovation in nuclear power for the first time in a while.
- [11:12]
Bel Lynn
Just to underscore the point that Steve made, power is really the primary constraint in developing more compute power for AI. And when you look at countries like China, which have the electrical capacity, the power to support compute capacity, it's really such a far cry from what the United States has in the current state of the electrical grid, which can't support, as it stands, the amount of electricity that will be needed to power AI. And so next year will bring a lot more potential innovation on the power side, whether in nuclear or maybe in geothermal even, and even more data center deals, for instance, because they're continuing to be built left and right. And also the Trump administration has signaled that it is more open to freeing up federal land for these data centers and also speeding, speeding up the permitting process. So in a sort of deregulated environment, it's likely that we'll see a lot more kind of that infrastructure being built up.
- [12:05]
Danny Lewis
Chip, I want to turn back to jobs for a second. How do you think this will impact entry level workers?
- [12:10]
Chip Cutter
We've seen this already be a weak spot where today's college grads are really struggling to find good jobs. They're struggling to find white collar roles because a number of companies have realized that as they roll out AI deeper into their organizations, a lot of the low level tasks, the sort of entry level jobs, the roles of analysts and others, can be done pretty well by a. I was talking to an executive. He made the point that these AI models make mistakes, but so do junior analysts. And actually this person had found that he could have kind of a tenth of his analyst team than he had at this point last year because he's just been able to roll out these tools that is an area to really watch. What happens if we have sort of a whole generation of workers who really struggle to get a foothold into the job market and to start to get on the traditional path to earning more and establishing their careers. The other, I think, is just looking at this era of the never ending corporate layoff, we're already starting to see signs of that. What we could see going forward is that as companies realize they're able to find these productivity gains, they're able to have software do more. You could just see the drip, drip, drip of layoffs happen throughout the year. It's not just that your company goes through one of these and then takes a deep breath and moves forward, but perhaps you're just continually going through this.
- [13:16]
Stephen Rosenbush
There's no easy answer for the job destruction or the weight that AI is putting on the job market. If someone loses their job today, the fact that their nephew or niece may have a fantastic new job down the road doesn't really help them right now. And you know, big question then is what happens to demand?
- [13:36]
Chip Cutter
That's it. Every time we write one of these stories, we get a lot of emails from readers who say, like, who's going to buy all these products if people don't have work? And that's a fair point. And I'm not sure we've really grappled with it.
- [13:46]
Stephen Rosenbush
I don't think we have.
- [13:47]
Danny Lewis
So that's a cheery note to head into a 2026 on Happy New Year.
- [13:53]
Chip Cutter
But it's also unpredictable. Discussion has shown this technology is moving so fast. So much has happened just this year that we probably wouldn't have anticipated if we had had this conversation in January.
- [14:02]
Stephen Rosenbush
Oh, absolutely. Bell and I participated in a roundtable I don't know, two months ago, and the mood was completely different. And like so much has happened, it's not just the views have changed, but so much has happened in the last two months.
- [14:18]
Bel Lynn
As maybe a counterpoint to the doom and gloom, there are cheerleaders in the tech industry and those who are really bullish on AI, who say that job displacement discussion and debate really skews too heavily on the job destruction side, that new jobs will be created. It's just a moment of change. And so there's also reason to be hopeful and excited that the new year will bring so many more new jobs like the Walmart agent builder position. But those require AI skills and familiarity with artificial intelligence as a whole, and that can be tricky for many to gain.
- [14:51]
Danny Lewis
Those were WSJ reporters Bel Lynn, Chip Cutter and Bureau Chief Stephen Rosenbush. And that's it for Tech News Briefing. Today's show was produced by Julie Chang. I'm your host Danny Lewis. Jessica Fenton and Michael Lavalle wrote our theme music. Our supervising producers are Melanie Roy and Katie Ferguson. Jessica Fenton is our technical manager. Our development producer is Ayesha Al Muslim. Chris Zinsley is the deputy editor and Falana Patterson is the Wall Street Journal's head of news audio. We'll be back later this morning. Morning with TNB Tech Minute. Thanks for listening.
- [15:30]
Bel Lynn
Amazon Business is 10 years old.
- [15:32]
Amazon Business Executive
We needed to build something that could address the unique purchasing challenges of different organizations. The need has never been larger and.
- [15:42]
Greater to use AI trying to manage.
- [15:44]
That and reduce complexity. It is a fundamental, you know, see change that is happening.
- [15:49]
We wanted to manage the number of suppliers we were able to timely and accurately get the supplies to the end user's possession. Amazon solved that problem for us.