Jerry Tworek, former VP of research at OpenAI, said the startup fumbled its lead over Google in the AI race.
OpenAI put the fear into not only Google but most of the tech industry with the release of ChatGPT.
Tworek said Google's recent gains with Gemini are as much about the search giant as they are about OpenAI's missteps.
Sometimes a comeback story starts with a fumble.
A former top OpenAI researcher said Google's AI renaissance is as much about OpenAI's missteps as it is about what the search giant got right.
"Personally, what I think you should consider Google's comeback, I think it's OpenAI's fumble," Jerry Tworek, a former VP of research at OpenAI, said on a Wednesday episode of Ashlee Vance's "Core Memory" podcast.
Tworek, who spent almost seven years at OpenAI, said earlier this month that he left the startup "to try to explore types of research that are hard to do at OpenAI."
OpenAI CEO Sam Altman declared a "Code Red" in December amid increasing competition from Google. The tech giant received wide praise across the industry for the capabilities of its Gemini 3 AI model, which some observers said had surpassed ChatGPT.
While declining to detail what he described as OpenAI's missteps, Tworek said that the pioneering AI company should never have lost the lead it established with the release of ChatGPT in 2022.
"If you are a company that is ahead and has all the advantages that OpenAI has you should always stay ahead," he said.
Overall, Tworek said, "Google did a lot of things right."
"Very clearly, Google started treating seriously at that moment, training large language models and, like, through OpenAI fumbling its lead, they are very, very close now in capability and in terms of models trained," he said, adding that the whole industry began to up its investment in AI when OpenAI showed ChatGPT could generate revenue.
As for OpenAI, Tworek said that the sheer toll of the AI race has led the non-profit-research lab-turned-public-benefit-corporation to place less of an emphasis on risky research that may not yield results. A spokesperson for OpenAI did not respond to Business Insider's request for comment.
"There are multiple aspects of certain things that are just hard to do in a company that has to compete in an extremely, extremely brutal and demanding race for having the best AI model in the world right now," he said. "One dynamic is there is naturally how much willingness of risks companies are willing to take from the perspective of trying to not fall behind."
Tworek said "all major AI companies" are facing pressure to show user growth and pay for GPUs while simultaneously competing to be the best available model.
"That does affect somehow your appetite for risk that you are willing to take," he said.
credit should read DON EMMERT/AFP via Getty Images
Nostalgic millennials are yearning for 2016 — from the Chicago Cub's World Series championship to skinny jeans.
But 2016 was also defined by political instability and corporate scandals.
Business Insider put together a quiz about the year's top stories to see how much you remember. Can you ace it?
How well do you remember 2016?
The now decade-old year is having a moment on Instagram and TikTok. Nostalgic millennials are pining for skintight jeans, blasting Justin Timberlake's "Can't Stop the Feeling," and lining up for the latest Jennifer Lawrence movie.
With the benefit of hindsight, 2016 feels like a calmer time. It was before a global pandemic, surging inflation, multiple wars, and the rise of AI and Big Tech.
But it didn't feel so simple back then.
Underneath the feel-good stories — like the Chicago Cub's ending their 108-year World Series drought or Hamilton's Tony Award win for Best Musical — the year was also defined by political upheaval, corporate scandals, and culture-shifting moments that still reverberate today.
So we put together a quiz to see how much you remember — including the good, the bad, and the political — from 2016.
5 AI and trade experts at Davos share their thoughts on the future of Robotics.
INA FASSBENDER / AFP via Getty Images
Digital twins are delivering real productivity gains for companies.
Labor shortages are accelerating the need for humanoid robots.
Tailored research and development is crucial for effective AI integration.
In a world of AI chatbots and agents, we're not talking enough about physical AI.
At a Thursday morning Davos panel moderated by Business Insider's Jamie Heller, robotics experts dived into ways bots are going to change the world — and their key limitations.
Here are three takeaways from the World Economic Forum panel, which featured robotics and automation experts and a trade minister:
1. Digital twins have moved from hype to real return
One phrase dominated the conversation: digital twins.
Executives from Siemens, Agility Robotics, and Automation Anywhere agreed that what once sounded like a distant possibility is now delivering real productivity gains for companies.
A digital twin is a digital replica of a physical object, person, or process. They can help organizations simulate real situations and their outcomes, allowing them to make better decisions.
"Digital twins have been around, but the technology has exponentially improved, and the computing power has exponentially improved," said Automation Anywhere CEO Mihir Shukla.
He added, "Let's say digital twins tell you that there is going to be a problem in manufacturing. You can proactively act on it. It has implications on your inventory. It has implications on your shipping planning."
Siemens' CEO Roland Busch said that the productivity boost is showing up in numbers.
He said that the company's output is 20% higher and that energy costs are 20% lower.
"You have much, much faster and ramping up, you don't make mistakes," if you use a digital twin, he said.
2. Labor shortages are driving humanoid robots
Robots can do work that humans are unable to — or don't want to — do, said Peggy Johnson, the CEO of Agility Robotics.
"It's very hard to hire for these manual jobs," she said. "They're dull and dirty, and they're dangerous at times because you're lifting over and over again, very repetitive. It's kind of mind-numbing work."
Faster progress in robotics means a better quality of life for the humans who are doing such jobs.
"The injuries are also another thing," she said about jobs that require a lot of heavy lifting. "We have an aging workforce. A lot of young kids don't want to go into environments like this. So the older employees are also getting hurt more."
Thani Ahmed Al Zeyoudi, the Minister of Foreign Trade for the United Arab Emirates, echoed that robotics was the future of "non-skilled" labor.
"It's too difficult to attract new young people to start working on the construction," he said. "Robotics is the main solution."
But building robots that are safe to use in human environments is also key, Johnson said.
3. You can't copy and paste R&D
Al Zeyoudi added that financing AI is only the first step toward AI integration.
"Sometimes you have the money, but you haven't diagnosed the right solution to the problem that you're having," he said.
He said that the UAE's push for AI includes government appointments, policy development, and research and development, which was the "main game changer."
"Many, many stakeholders are avoiding that because they say someone else will do it," he said, referring to R&D. "I'll just copy and paste it."
"But when it comes to AI, it's too late," Al Zeyoudi added. "You have to start working on R&D because you ensure to customize things to your own ecosystem, your own environments, your own conditions."
Gavin Newsom went hard with anti-Trump messaging during his session at Davos.
He said the US was not living under the rule of law but rather "the Rule of Don."
To illustrate his point, he waved around kneepads for corporate leaders who he said were kneeling to Trump.
California Gov. Gavin Newsom used his 30-minute speech at Davos to fire salvo after salvo at President Donald Trump.
About 10 minutes into his interview with Semafor's Ben Smith, Newsom said that the US was living under "the rule of Don."
"Co-equal branches of government, the rule of law, popular sovereignty," he said. "Tell me that that reflects the America you read about today."
"There's no rule of law. It's the rule of Don," he added.
Newsom accused the US Congress of being "supine" and called out American universities, law firms, and corporate leaders for selling out to the administration. To drive the point home, he also brought in bright red knee pads for executives — though he didn't name any CEOs.
"There are knee pads that are available to purchase. The last round of knee pads sold out just as our law firms are selling out," he said.
The governor made the comments on the fourth day of the World Economic Forum.
Newsom, at several points, launched into passionate, debate-like monologues, which elicited rounds of applause from the audience.
The politician is a Davos regular. He attended the forum multiple times while mayor of San Francisco from 2004 to 2011.
Political experts told Business Insider that his return to the event sets the stage for a potential 2028 presidential run. He announced last week that the top of his Davos agenda was to rebut Trump's vision for the future of capitalism.
Trump called out Newsom in his speech on Wednesday, saying he knew that the governor was at the event. He called Newsom a "good guy" and said he used to "get along so great" with him.
Newsom ran into some trouble in Davos, too. He was denied entry to the USA House, despite being scheduled to speak at a fireside chat.
Newsom criticized the administration in a post on X, writing: "How weak and pathetic do you have to be to be this scared of a fireside chat?"
Newsom was offered an invitation to the USA House for a "VIP Nightcap" several hours after Trump's speech, which he declined.
On Tuesday and Wednesday, Newsom also exchanged barbs with US Treasury Secretary Scott Bessent. At a press conference on Wednesday, Bessent called Newsom "smug, too self-absorbed, and too economically illiterate to know anything."
It's not just Trump: Newsom has spent the last several years being a foil to major Republicans. He has a history of going toe-to-toe with Florida Gov. Ron DeSantis, who was once Trump's biggest rival for the GOP presidential nomination.
Mykhailo Fedorov, Ukraine's new defense minister, has set an audacious new goal for his forces: killing 50,000 Russian troops per month.
Andrii Nesterenko/Global Images Ukraine via Getty Images
Mykhailo Fedorov outlined a brutal new goal for his troops: Kill 50,000 Russian soldiers every month.
The defense minister said Ukrainians are already reporting 35,000 kills a month, confirmed by video.
It comes amid heightened talk that Russia may soon no longer be able to sustain its supply of troops.
Ukraine's new defense minister has set a new goal for its forces to kill 50,000 Russian troops a month, a sharp increase from the losses that Kyiv says it's inflicting on the Kremlin now.
In a meeting with reporters on Tuesday, Mykhailo Fedorov said the new figure would be a core "strategic objective" for Ukraine.
"The objective is to impose costs on Russia that it cannot bear. In this way to force peace through strength," Fedorov said.
Notably, he specified only battlefield fatalities.
The new standard comes as Russia intensified its offensive in some areas of the front lines last year, aiming to take key cities such as Pokrovsk. The nature of its operations — often reliant on continuous high-casualty infantry attacks — has led to political discussion in recent months that the Kremlin may soon reach a point where it will lose more soldiers than it can recruit.
Ukraine's commander in chief, Gen. Oleksandr Syrskyi, said in December that his forces had, for the first time, killed and wounded more Russian soldiers in one month than the Kremlin had called up for the same time period.
Fedorov, who was confirmed in his role last week, said on Tuesday that Ukrainian troops are already documenting 35,000 kills every month via video.
NATO Secretary General Mark Rutte cited another figure while speaking in Brussels on January 16, saying that 20,000 to 25,000 Russian troops were dying every month.
While differing significantly, both figures indicate that Russia is fighting the war at a high cost.
Rutte said the death rate has become "unsustainable" for Russia, and drew comparisons to the Soviet Union's losses during the 1979 invasion of Afghanistan, when an estimated 15,000 Soviet troops died every month over nine years.
Likewise, Fedorov expressed hopes that ramping up a focus on lethality will exhaust Russia's war machine.
"If we reach the figure of 50,000, we will see what happens to the enemy. They treat people as a resource, and problems with that resource are already obvious," he said.
Russian soldiers train in Rostov-on-Don in mid-January.
Arkadii Budnitskii/Anadolu via Getty Images
Ukraine hopes that Russia will run out of troops
Russia does not disclose how many of its troops are wounded or killed during the war. However, international and Ukrainian analyses say the number is likely tracking past 1 million people over nearly four years of war.
The UK's defense ministry said in early January that the Kremlin may have suffered at least 1.2 million total casualties, which includes both wounded and deceased, since February 2022.
Meanwhile, the Kremlin has relied largely on hefty sign-up bonuses to attract local contract soldiers, though it's also been found to sustain troop levels through other means, such as by recruiting from prisons or hiring foreigners via informal channels.
Some of those strategies may be waning. On Tuesday, the Russian independent media outlet Vertska reported that contract soldier sign-ups were slowing in some areas of the country.
Data from an unnamed source in the Moscow mayor's office, it wrote, showed that the capital had seen a roughly 25% annual drop in recruitment in 2025.
It's unclear how Fedorov intends to specifically target increasing combat kills. He said in his opening remarks to Ukrainian lawmakers last week that he intended to divert more staffing to drone units. Drones are thought to be responsible for roughly 70 to 90% of casualties inflicted during the war.
His new objective also comes against the backdrop of the US attempting to negotiate a ceasefire between Ukraine and Russia. Many of the Kremlin's latest war actions, such as attempts to take Pokrovsk, have been seen in Ukraine as efforts to posture into a position of strength.
At 32, Fedorov is Ukraine's youngest-ever defense minister. He's begun his new appointment with promises of aggressive reform of the military's structure and how it distributes resources to its fighters.
Fedorov was previously Ukraine's minister for digital transformation.
The interview on the "Relentless" podcast covered dozens of topics, from the internal company culture and work schedule to the eyebrow-raising way Elon Musk's AI company builds its data centers.
Elon Musk's companies are famously wary of the press and media. And while it's not clear whether Ghori's exit is related to the podcast interview — neither xAI nor Musk commented on the former employee's quotes or departure when contacted by Business Insider — some big names like MrBeast are speculating as much. Ghori hasn't commented publicly about the circumstances of his departure and did not respond when contacted by Business Insider.
So what exactly did the now-former xAI employee talk about?
Read on for 10 of the most interesting things Ghori said on the podcast.
1. XAI's data centers are built on temporary 'carnival' leases, Ghori said
How is xAI building its data centers so quickly? Through temporary licenses, Ghori said.
"It was the fastest way to get the permitting through and actually start building things," Ghori said. "I assume that it will be permanent at some point."
Ghori said that the temporary leases were an exception granted by the local government, one made for carnivals. The host, Ti Morse, laughed: "So xAI is actually just a carnival company?"
"It's a carnival company," Ghori responded.
2. XAI is full of 'very good' AI employees — but that can lead to org chart confusion
AI visionaries often talk about a world where managers run a team of agents, not employees. They seem to be there already at xAI.
The company is rebuilding its core production APIs, Ghori said. The team leading it is one person and 20 agents. "They're very good, and they're capable of doing it," he said.
At another point in the podcast, Ghori described the confusion that AI employees can cause.
"Multiple times I've gotten a ping saying: 'Hey, this guy on the org chart reports to you. Is he not in today or something?'" he said. "It's an AI. It's a virtual employee."
XAI teams are kept small, even without AI employees. The iOS team had three employees at the time of the Grok Imagine launch, Ghori said. He was the third.
3. Each 'commit' to the xAI code base is worth around $2.5 million
How valuable is each commit to xAI's repository? They did the math, Ghori said: It's $2.5 million.
"I did five today," Ghori said. His work for the day would be valued at $12.5 million.
4. 'When we pick up new products from Nvidia or whoever, not everything works'
One of Elon Musk's key roles at xAI is as a fixer.
Ghori said that, when the company picks up new products from the likes of Nvidia, not everything works. That's when Musk gets on the phone, Ghori said.
"We would work side-by-side until that was resolved," Ghori said. "Otherwise it would have taken weeks of back-and-forth."
5. Musk's Cybertruck incentive
The xAI CEO made an unusual offer when xAI's engineers were setting up new GPU racks.
"Elon's like, 'OK, you can get a Cybertruck tonight if you can get a training run on these GPUs in 24 hours,'" Ghori said.
The engineer — whom Ghori only referred to by their first name, Tyler — won the bet. Now, Ghori said he sees Tyler's Cybertruck outside his lunch window.
6. Nobody told him what to do while onboarding, Ghori said
The teams within xAI are limited and blurry, Ghori said.
That made onboarding a challenge, he added, as nobody told him what to do.
"My first day, they just gave me a laptop and a badge," Ghori said, adding that he wasn't assigned a desk.
Ghori sought out cofounder Greg Yang, who had been instrumental in his hiring. He soon started working on the Ask Grok feature in X.
7. Sleeping pods, bunk beds, and tents.
Elon Musk's companies have a long history of overnighting at the office. Former Twitter director Esther Crawford generated headlines when she posted a "cheeky" photo of herself sleeping at the company's headquarters.
XAI seems to have embraced this reputation. The company has sleeping pods and bunk beds, Ghori said.
"When the tent picture came out, everyone kept sending it to me," Ghori said. "We have tents, but I've never seen that many out at once."
Who responds when Musk spots a late-night problem with X? "Whoever is awake," he said.
8. XAI's big 'human emulator' bet
Ghori worked on the Macrohard team — a tongue-in-cheek play on the opposite of Microsoft — that is developing "human emulators."
The xAI engineer explained the concept in reference to Optimus, Tesla's humanoid robot. Just as Optimus performs physical human actions, these emulators will perform digital human actions.
The emulators will do anything that a human needs to look at a screen, use a keyboard and mouse, and make decisions, Ghori said.
9. Dormant Teslas could power the human emulators
XAI wants to roll out the human emulators slowly, then all at once, Ghori said. The goal is to scale to one million emulators.
There are 4 million Tesla cars in North America alone, Ghori said. They're sitting idle for 70-80% of the time, he said. Why not pay owners to lease time off their cars and run the emulator on them?
"That's something without any build-out requirement," Ghori said.
This isn't the first time using dormant Teslas to power new ambitions has been mentioned. Elon Musk said at Tesla's November shareholder meeting that the vehicles could offer a "massive distributed AI inference fleet."
10. XAI's new models are planned out far in advance
Consumers can currently use the Grok 4 model, which xAI released in July.
XAI is working far ahead, Ghori said. He joined in March 2025, according to his LinkedIn profile. Grok-5 was planned even before he joined, Ghori said.
The model was "planned out and designed" far in advance, he said.
A top BYD executive says countries are slowing the EV transition by changing policy too often.
Asad/Xinhua via Getty Images
Policy changes are slowing the global EV transition, a top BYD exec says.
Countries that go "back and forth" on EV policy risk confusing manufacturers, Li said.
When governments give a "very clear line," automakers can focus on execution, Li said.
The global EV transition has a policy problem, says BYD executive vice president Stella Li.
Li said in a panel session on EVs on the second day at Davos that governments keep changing the rules about EVs. The stop-start approach makes it harder for automakers to commit capital, plan product cycles, or build supply chains with confidence, even as competition across the EV markets in China, the US, and Europe heats up.
Li said that when countries go "back and forth" on their EV policy, it creates a pattern that "will confuse manufacturers."
In contrast, when governments give a "very clear line," automakers can focus on execution, Li said.
Akaash Vishal Hazarika shares how AI is reshaping the skills required for software engineering jobs.
Tech companies now expect engineers to integrate AI into coding, debugging, and workflows.
He suggests candidates combine core engineering skills with AI expertise for career success.
This as-told-to essay is based on a conversation with Akaash Vishal Hazarika, a 29-year-old senior software engineer based in Seattle. It's been edited for length and clarity.
I've been a Big Tech employee for the past eight years, working at Google, Amazon, Splunk, and now Salesforce. I've had a front-row seat to witness the changes in the tech landscape.
I've learned which skill sets software engineers need to land a job offer in the AI era. Tech companies agree that AI makes engineers more productive, so engineers are expected to use it to build things more quickly and reliably.
I personally make heavy use of AI to help me with boilerplate stuff so that I can concentrate on the hard stuff, like system design and complex business logic.
Preparing to be a software engineer isn't the same as it was five years ago
When I was interviewing for software engineering jobs in 2020, LeetCode and system design were the de facto standards for cracking a job interview. Job seekers who had an advanced understanding of data structures and algorithms would come out on top.
Today, this is just the baseline expectation. AI is now widely used for coding, review, and design, so tech companies, especially startups, expect more from candidates.
Some skills remain the same: Software engineers still need a problem-solving mindset and should know how to scale systems and leverage cloud services. New hires must now also understand prompt engineering — how to leverage AI to code a solution more efficiently.
You need to be able to use AI for error handling and bug fixing. AI systems integration is another important new skill that involves incorporating AI into existing workflows, scaling AI systems, and determining when to use traditional versus AI solutions for business decisions.
Interviews have also evolved with the arrival of AI
You're still expected to have fundamental knowledge of core system design, data structures, and algorithms. You can still expect interviewers to test your problem-solving approaches, and if you know how to make the correct tradeoffs in time and space. Interviewers still care about debugging skills, since AI makes a lot of fundamental logic errors.
You also need to be prepared for some new things. I've seen firsthand that working with AI assistants in live screen-sharing mode is now allowed in some interviews. Interviewers are looking to see if you can help them achieve their business outcomes by combining software engineering skills with AI prompting.
In an interview I had with a Silicon Valley startup in 2024, I expected the hiring team to give traditional coding challenges. I was taken aback when I was given a huge code file and asked to debug a buggy behavior, and the interviewers explicitly said I could use AI assistance.
I ignored the invitation to use AI, thinking I was supposed to do it myself, and ended up spending a lot of time on the problem to no avail. I failed that interview. That was an eye-opener for me about AI's new role in this field.
You may also be asked system design questions about where AI should be integrated into the current business workflow, or to discuss the trade-offs of using AI and traditional approaches in various problem contexts.
During my own system design interviews, I noticed companies starting to ask questions about how to integrate AI to enhance existing systems and support a particular model lifecycle in terms of infrastructure. What I've seen is that companies give access to a small codebase and expect you to deliver a small feature in about one hour, which is impossible without AI. With AI, you can roll it out easily.
I've also been asked behavioral questions, like "How do you plan to evaluate the use of AI in order to make the workflow better?" and "How do you plan to balance automation and manual oversight?"
For new graduates, here's how to show hiring teams that you're equipped for the emerging needs of the software engineering role.
1. Cultivate a production mindset
Start by contributing to open-source projects on AI or any other GitHub project. This demonstrates that you can navigate a production codebase and work on it independently to build a new feature or fix a bug. For solo repositories, include a README that explains the rationale for your decision.
2. Build a portfolio of AI-integrated projects
Incorporate the use of AI in traditional repositories to demonstrate hands-on experience with AI integration. Don't just deploy and run it locally — try to deploy it to the cloud. Many cloud providers provide free student credits.
3. Master cloud tooling and AI prompting
Focus on prompting AI to drive intended outcomes more effectively by providing structured input and output. Skills like AWS or GCP cloud certifications demonstrate your ability to be a keen learner.
4. Practice Leetcode-styled questions
Learn different problem-solving patterns and practice these skills regularly. Muscle memory and pattern recognition go a long way.
For experienced software engineers, your biggest asset is your deep engineering experience — here are three tips that will help you in 2026 and beyond.
1. Map your specialty with complementary AI skills
Pairing years of engineering expertise with impressive AI skills will get the attention of hiring managers. Consider boning up on these complementary skills:
Backend: Focus more on tasks related to scaling systems, especially AI, such as managing throughput and latency during deployment and maintaining versioning.
Data engineering: Build knowledge and skills in Kubeflow, MLFlow, Apache Spark, and Kinesis — these skills are becoming increasingly important.
Site reliability engineering: Learn about SRE, which involves tracking AI usage and cost, and building fallback mechanisms when models misbehave in production.
2. Develop an AI product mindset
Senior software engineers don't just build; they also need to strategize and innovate. Try to understand the trade-offs of relying on a third-party API versus using an open-source one and fine-tuning it for your business purposes. Questions on cost, reliability, and maintainability should always be top of mind as you develop an AI product mindset.
3. Make use of AI in your current workplace
Identify workflows that have a great deal of manual involvement in your current job, and use AI to see if you can improve them. You can leverage AI to help you brainstorm on how to make it more efficient.
Both veteran and newbie engineers should stay curious: find out the cool technologies and AI agents available in today's market, but don't abandon fundamental engineering approaches. The hybrid understanding of these technologies is what makes you valuable.
If I were interviewing, I'd position myself as a 'hybrid engineer.' Don't just be a pure coder or just a prompt engineer. Be the bridge.
"There's enormous issues for our economy and for our democracy that have got to be dealt with," Sanders said. "And I feel we're not ready to do that."
Nathan Posner/Anadolu via Getty Images
Bernie Sanders has called for a nationwide moratorium on AI data center construction.
He told Business Insider last week that he plans to introduce legislation on the issue soon.
Few progressives have lined up behind Sanders's call, but concerns about data centers are growing.
Sen. Bernie Sanders wants a nationwide pause on AI data center construction. Few of his colleagues are willing to go there yet.
"In terms of the actual policy prescription, it's something that I haven't made a determination yet on," Democratic Rep. Alexandria Ocasio-Cortez of New York told Business Insider.
The Vermont senator and two-time presidential candidate first made the call in December, saying in a video posted to X that a national moratorium would "give democracy a chance to catch up with the transformative changes" that the technology is bringing.
Sanders told Business Insider last week that he plans to introduce legislation to back up his call soon.
"There's enormous issues for our economy and for our democracy that have got to be dealt with," Sanders said. "And I feel we're not ready to do that."
But his proposal is unlikely to become reality, given GOP control of Washington.
"I mean, this Congress isn't going to do that," Democratic Rep. Mark Pocan of Wisconsin told Business Insider. "He's pointing out the right problems."
Republicans have been generally dismissive of Sanders's idea. Even Sen. Josh Hawley of Missouri, a major critic of the AI industry, signaled he wouldn't go as far.
"If these AI companies want to build these data centers and local folks want to give them permits for it, I mean, that's up to local voters," Hawley told Business Insider.
A handful of Sanders's progressive allies are backing him up. Rep. Ilhan Omar of Minnesota told Business Insider that a moratorium "would be a good idea," while Rep. Rashida Tlaib of Michigan issued a full-throated endorsement in a post on X.
"I fully support the call for a national AI data center moratorium," Tlaib wrote.
And while few Democrats have fully endorsed the idea, it's spurring a conversation within the party about the impact of data centers and AI.
"I don't want a moratorium on data centers but Bernie is right that we are sleepwalking past risks which alarm even the optimists," Democratic Sen. Brian Schatz of Hawaii wrote on X last month.
"What we do know is that these AI data centers are just uncontrollably jacking up people's energy costs," Ocasio-Cortez told Business Insider. "There are a lot of problems that arise from these data centers, and I think that they certainly should not be getting the blank check from Congress."
By contrast, the Trump administration has been broadly supportive of the AI industry and the associated data center boom, including taking steps to limit states' ability to enact AI regulation.
"Bernie makes clear that the debate over AI is not about states rights or affordability," White House Crypto and AI Czar David Sacks wrote on X in December. "He would block new data centers even if states want them & they generate their own power. It's about stopping progress completely so China wins the AI race."
Yet even the administration has come to recognize that data centers are engendering local resistance, in large part due to rising electricity costs.
Last week, President Donald Trump declared that he wanted Big Tech firms to "pay their own way" on data centers, with AI companies footing more of the bill for the electricity their facilities consume.
"I never want Americans to pay higher Electricity bills because of Data Centers," Trump wrote on Truth Social.
And on Friday, the administration and a group of governors called on a major power grid operator to hold a new emergency power auction amid rising costs driven by AI data centers.
The cities building the most homes rely more heavily on immigrant labor.
Allen J. Schaben / Los Angeles Times
The cities that build and remodel the most homes rely the most heavily on immigrant workers.
Dallas-Fort Worth led the US in building permits. Its construction workforce was 61% foreign-born.
Mass deportations and restrictions on immigration threaten to deepen the worker shortage.
The American homebuilding industry relies heavily on immigrant workers. That's especially true in the cities that build and remodel the most homes, according to new research from Harvard's Joint Center for Housing Studies.
In the seven metro areas that issued at least 150,000 residential building permits between 2019 and 2023, an average of 54% of construction trades workers were foreign-born, the Harvard report found. The metros building and remodeling the most homes — from Los Angeles and Washington, DC, to Dallas and Houston — rely on a workforce that's often more than 60% foreign-born.
The construction industry faces a nationwide worker shortage in the hundreds of thousands. Given its reliance on foreign-born workers, President Donald Trump's mass deportations and restrictions on immigration threaten to deepen the worker shortfall, said Anirban Basu, the chief economist at Associated Builders and Contractors, a trade group that endorsed Trump in 2024.
As a result, economists and housing researchers expect the most dynamic US housing markets will be hit the hardest by rising construction costs — driven by higher labor costs and delays, in part due to the worker shortage.
"These places that are most reliant on immigrant labor are going to feel those effects most acutely, and may then have a hampered ability to respond to housing supply and demand needs," said Riordan Frost, a senior research analyst at the Harvard Center and the author of the report.
Metros that build fewer homes tend to have far fewer immigrants as a share of their construction workforce. In metros that granted between 75,000 and 149,999 permits, an average of 40% of the workers were foreign-born. And in metros that permitted less than 75,000 homes, 22% of the workforce was foreign-born. Still, immigrants made up a disproportionate share of the construction workforce in those places, too.
The Dallas-Fort Worth-Arlington metro area led the country in issuing new building permits between 2019 and 2023. During that period, 61% of the area's construction workers were immigrants. In that same period, nearly three-quarters of construction workers in the Miami-Fort Lauderdale-Pompano Beach area were foreign-born.
"There's demand for labor in these places because there's so much homebuilding activity, and that is what creates the economic opportunity for immigrants to come in and fill these positions, especially if they're positions that native born people aren't as likely to work in," Frost said.
A recent survey by the Associated General Contractors of America found that 28% of construction firms say they've been affected, either directly or indirectly, by immigration enforcement. More than 90% of all firms that were hiring said they were having trouble filling open roles, and 45% of all firms said they experienced project delays because of a shortage of workers.
"There's no question in my mind the stepped-up immigration enforcement is serving to drive up construction delivery costs," Basu said. "If all of a sudden these communities are no longer able to supply as much new housing, then their economic growth will tend to stagnate."
At the same time as the administration is cracking down on legal and illegal immigration, it's not doing enough to boost domestic construction training, AGC said.
The Trump administration says it's working with employers to streamline visa applications for temporary workers and boost vocational training.
"There is no shortage of American minds and hands to grow our labor force, and President Trump's agenda to create jobs for American workers represents this Administration's commitment to capitalizing on that untapped potential while delivering on our mandate to enforce our immigration laws," White House spokeswoman Abigail Jackson said in a statement.
Immigrants have long made up a disproportionate share of the country's construction tradespeople, particularly in homebuilding. And, as of 2024, an estimated 15 to 23% of the total construction workforce was living in the country illegally. In recent years, the number of native-born construction workers has shrunk significantly.
In the long term, improvements in construction productivity — including the growth of modular, factory-built housing — and a growing number of women and young people entering the industry could ease the worker shortage.
But for the next several years, at least, the worker shortage is set to worsen, helping drive up building costs and home prices.
"The construction workforce is set to become more expensive going forward," Basu said. "And then when you layer on top of that, of course, increases in healthcare costs, electricity rates, those kinds of things, there's even more pressure on these households going forward."
When you buy through our links, Business Insider may earn an affiliate commission. Learn more
Casper coupons can help save you hundreds of dollars on a bed.
Casper
Buying a new mattress doesn't have to be complicated, and Casper takes the stress out of the process. The popular bed-in-a-box brand offers thoughtfully designed mattresses that work for a wide range of sleep styles and budgets, all shipped straight to your doorstep. And with the right Casper coupon, you can save hundreds on your next mattress purchase.
Our mattress experts have tested eight of the modern Casper models, and each one stood out for its comfort, support, and durability over time. The Casper Snow remains a top choice for hot sleepers thanks to its cooling design and breathable materials. For a closer look at how each mattress performed in testing, be sure to read our full Casper mattress review.
Casper often runs discounts on select mattresses, but the best deals come from stacking those offers with a Casper promo code. Some promo codes offer an additional 25% off, resulting in major savings on higher-end models. To help you get the most value possible, we've rounded up the best Casper coupons and promo codes below so you can upgrade your sleep setup for less.
Today's best Casper coupons and promo codes
Previous Casper promo codes
Are there any other Casper mattress sales running at the moment?
You can find a full list of clearance markdowns on Casper's website anytime. As of the time of writing, Casper is running the following promotions.
Use Casper coupons on these items
You can use Casper coupons to save extra on Casper mattresses featured in our guides. The queen-sized Casper One mattress is already on sale for $799, down from the original $999. But when you use code PERKSPOT15, you can save an extra 15%, lowering the price to just $679.15. Similarly, the queen-sized Casper Snow mattress, one of our favorites for hot sleepers, is also on sale for $2,075 instead of the original $2,595. But when you apply the promo code NEWYEAR26, you can lower it by another 25% to $1,556, helping you save over $1,000 in total.
For the latest buying advice updates, follow us onInstagramandWhatsApp
How to use Casper coupons and promo codes
Casper coupons are pretty easy to use. Copy the code you're planning to apply. Add the product you're buying to your cart, then click "Apply promo code" to open the entry field. Paste your discount code there, then click Apply.
Be sure to apply your promo code to your purchase by clicking "Apply promo code" before completing your Casper checkout.
Casper
Can you stack Casper coupon codes?
Casper coupon codes are not stackable, meaning you can only apply one to each purchase. Therefore, it's important to use the best one available.
Luckily, the coupon codes work with any ongoing Casper sales.
Can I use Casper coupons in-store?
Casper coupon codes vary in whether they can be used in-store. Most often, they are limited to online use, but in-store associates can price-match any discounts you have proof of.
What mattress should I buy from Casper?
Casper makes a whole lineup of great beds designed for different types of sleepers. You can learn more in our full Casper mattress guide, or check out these favorites of ours:
Casper
Casper coupons: frequently asked questions
Does Casper offer discounts for teachers, students, or the military?
Casper offers exclusive military, education, and healthcare discounts for verified members. Active-duty service members, veterans, military family members, teachers, students, nurses, healthcare professionals, first responders, and hospital employees can all save 5% on Casper purchases.
Casper offers a 10-year limited warranty that automatically comes with mattresses purchased on Casper.com. It covers defects, including abnormal deterioration, physical flaws in the foam, and manufacturing errors in the zipper assembly. If you bought your bed from another retailer, you can register your bed here.
Final sale and clearance items receive a shorter warranty than other purchases.
What is Casper's returns/refunds policy?
Casper's return policy aims to be as painless as possible, offering free returns anytime after the 30-night adjustment period. Just reach out to support within the first 100 nights of your purchase to have the brand remove the mattress from your home and provide a full refund.
Other products, like pillows, sheets, and accessories, can be returned by printing out a given shipping label and sending it back for free.
Does Casper offer free shipping?
Every mattress order with Casper receives free, no-contact delivery. As the term "bed-in-a-box" suggests, each mattress arrives tightly coiled in a box for you to place on your bed and unfurl.
Other items, such as headboards and bed frames, may incur a shipping fee depending on where you live. Learn more about the specific costs here.
Does Casper offer white glove delivery?
Casper offers an in-home delivery and setup service starting at $199. After placing your order, you can schedule a delivery date for two to four weeks in the future. From there, you'll receive a four-hour delivery window the day before to prepare for your new product's arrival.
Will Casper remove my old mattress?
If your order includes in-home delivery and setup, Casper will offer disposal options for your old mattress. If your order does not include in-home delivery, check Casper's website or contact their customer support team for suggestions.
When is the best time to buy a mattress?
Mattress retailers typically offer the best sales and deals during Presidents Day, Memorial Day, Black Friday, and Cyber Monday. Shopping during those times, combined with one of our coupon codes, will ensure you get the best price.
Does Casper only offer mattresses?
In addition to mattresses, you can shop for pillows, bedding, and furniture on Casper's site. You can even purchase dog beds from Casper to show your pets some love.
How does Casper compare to other online mattress companies?
Many online bed-in-a-box companies have risen in popularity in recent years, and each has its own merits and drawbacks. Check out our full guide to the best mattress to see how Casper compares to competitors in criteria like price range, types of mattresses, and delivery offerings.
Are Casper mattresses designed for specific sleep types and positions?
Casper designs their mattresses to provide tailored support no matter which position you sleep in, so you should be covered whether you prefer to sleep on your back, side, or stomach. The brand's patented "Zone Support" technology works to keep your spine aligned in any position for maximum support.
Can I reuse a Casper coupon?
For the most part, Casper promo codes can only be used once per person. However, there may be exceptions, so be sure to check the terms and conditions for the specific code you want to use.
To lose fat without losing muscle, you need to eat a high-protein diet while staying in a gentle calorie deficit, trainers told Business Insider.
Cavan Images/Getty Images/Cavan Images RF
When we say we want to lose weight, we typically mean shedding fat but not muscle.
Muscle helps us to look "toned" and supports our metabolism.
To lose fat without losing muscle, eat enough protein and strength train regularly, two top trainers said.
If you want to lose fat without losing muscle, there are three things you need to know.
That's according to Magnus Lygdback, a Hollywood personal trainer who helped Alexander Skarsgård prepare to play a Viking warrior in "The Northman," and Harry Cox, a personal trainer to busy C-suite execs based in London.
To lose weight, you need to eat fewer calories than you use, but if that number drops too low, you'll start losing muscle along with fat.
"Because your body is in an energy deficit, it's going to try and find energy wherever it can, and sometimes that ends up being muscle," Cox told Business Insider.
You may look slimmer, but it could backfire in the long run, because muscle mass is essential for overall health, he said. It plays a key role in regulating metabolism, including glucose levels, and increases the number of calories your body uses at rest. All this impacts body composition.
We also naturally start to lose muscle with age, and building and maintaining it allows us to move around freely, lift things, and prevent injury longterm.
Both trainers endorsed establishing long-term healthy habits like eating a balanced diet and getting enough sleep over quick fixes, which can be structured around the following three tips.
Harry Cox trains C-suite execs in London.
Club Q
1) Strength train with progressive overload at least twice a week
Strength training regularly, particularly if you're trying to lose fat, will help you preserve muscle mass, Cox said. It signals to the body that we need it and must keep it, he said.
The most effective and evidence-backed way to build muscle is to train with progressive overload. This is where you work your muscles to the point of failure — where you could barely do even one more rep— and continually add more weight or reps as your body adapts over time.
Cox said the same principle applies to maintaining muscle. "If you're just going to the gym and lifting some of the lightest dumbbells and hope that's going to help you preserve muscle, that's not the case. The training needs to be somewhat intense," he said.
He said to aim for a minimum of two strength training sessions a week, but that you'll likely get better results from three or four.
Each workout should include at least one pulling exercise — such as a chin-up, pull-up, or lat pull-down — a push exercise — such as a push-up, bench press, or shoulder press — and some kind of leg exercise — such as a lunge, squat, or deadlift.
"If you start putting in place those healthier habits of eating more protein, exercising more frequently, a lot of the time, the fat loss happens as a byproduct without even focusing on it," Cox said.
2) Eat enough protein
Learn which foods are good sources of protein.
bhofack2/Getty Images
Protein is the building block of muscle, so eating enough is crucial for maintaining muscle while losing fat, Lygdback said.
Whether or not you're trying to lose weight, you should aim to eat the same quantity of protein to maintain your muscles, he said.
"You don't really want to adjust that too much if you ask me. And I never do," he said.
The scientificliterature recommends eating between 1.6 and 2.2 grams of protein per kilogram of body weight a day to support recovery and maintain or build muscle mass, Cox said.
A person who weighs 60 kilograms would aim to eat 120 grams of protein each day, for example.
If you're struggling to eat enough protein, the trainers recommended educating yourself on which foods are high in protein, using protein powder if necessary, and not being afraid to order double protein in a restaurant.
3) Stay in a gentle calorie deficit
To lose weight, you need to burn more calories than you eat, known as a calorie deficit. But building muscle requires being in a calorie surplus. That's why it can be hard to lose weight without shedding muscle instead of fat.
You can achieve it though by tracking macronutrients (protein, fats, and carbohydrates), Lygdback said . You don't have to weigh out your food at every meal, but it might be helpful to track your meals for a few days to get a sense of what 30 grams or protein, or 500 calories of rice looks like, for example, he said.
Chicken, rice, and broccoli is the staple gym bro diet. But you don't need to eat that to reach your goals, he said. Just make sure you have enough protein, while eating in an overall calorie deficit. "You want to make sure to keep the protein high enough, and the carbs and fat low enough," he said.
Lygdback recommended cutting your daily calories by 10 to 20%. "You don't want to go too extreme," he said.
Business Insider previously reported on how to calculate your macros depending on your weight and activity levels.
Companies rushing to adopt the same AI tools risk losing their competitive edge, independence, and long-term resilience, a think tank CEO says.
APHOTOGRAFIA/Getty Images
The CEO of a digital economy think tank said relying on the same AI tools kills competitive edge.
Mehdi Paryavi said firms replacing people with subscriptions risk dependency.
Using identical AI tools can drain innovation as competitors end up buying the same brain, he said.
As companies rush to adopt AI to boost productivity and cut costs, they may be setting themselves up for a new problem: losing what makes them different.
Mehdi Paryavi, CEO of the International Data Center Authority, said widespread reliance on the same AI tools risks flattening competitive advantage across industries, because firms increasingly rely on identical systems to think, write, and decide for them.
Paryavi said that as AI tools become cheaper, more powerful, and more widely deployed, companies risk outsourcing the very thinking that once differentiated them.
While AI can boost efficiency in the short term, he said, relying on shared models and standardized systems could leave businesses competing on cost and speed alone — eroding originality, strategic depth, and long-term advantage.
"If you and your competitor are all using the same service, you have no edge over each other," Paryavi told Business Insider.
"Their AI and your AI against each other — I don't know who's going to win."
When everyone uses the same brain
As generative AI becomes embedded across workplaces, Paryavi warned that the biggest risk isn't automation — it's uniformity.
When companies rely on the same large language models trained on the same data, decision-making, writing, and problem-solving can start to converge, shrinking the space for creative divergence.
That concern echoes warnings from researchers and academics who say AI can produce polished output at scale, but also flips human thinking by delivering fluent answers before understanding, creating an illusion of expertise that weakens judgment and depth.
When everyone relies on the same models trained on the same data, Paryavi said, creative divergence shrinks.
"The beauty of our world is that we have different choices because we think differently," he said. "That's where innovation comes from."
Efficiency today, dependence tomorrow
It's not just a question of companies all thinking the same — Paryavi warned that treating AI as a shortcut to efficiency can quietly hollow out human judgment, expertise, and control, leaving businesses faster in the short term but more fragile over time.
Over time, Paryavi said, that shift can erode internal expertise and decision-making capacity.
"What they don't think about is that initially it might sound more efficient and more productive and cheaper," he said. "But this is going to be very expensive down the line."
One risk, Paryavi said, is dependency. As firms replace employees with AI subscriptions, they become increasingly reliant on external vendors to function effectively.
Paryavi compared the AI boom to the early 2000s rush to cloud computing, when many companies initially adopted third-party infrastructure but later repatriated workloads in-house as costs, complexity, and vendor lock-in became concerns — a trend commonly referred to in tech as cloud repatriation.
The same dynamic could play out with AI, Paryavi said — except with even higher stakes. As companies downsize human teams, they also lose institutional knowledge and the ability to operate without automation, he said.
"You've killed all your chances of ever becoming independent as an organization," he said. "You've fired your manpower. You've made them no good."
AI, he said, is not inherently harmful. In fields such as medicine, scientific research, and disaster prediction, it can significantly accelerate progress.
But without clear guardrails, companies risk trading long-term resilience for short-term speed.
"It's a very powerful tool," Paryavi said, comparing AI to an atomic bomb. "If that [an atomic bomb] can eliminate an entire population physically, this [AI] can eliminate humanity cognitively."
