Meta CEO Mark Zuckerberg says today’s startup founders have a unique edge — access to powerful AI tools that didn’t exist when he was building Facebook.

His remarks, made during a session at the Stripe Sessions conference this week, underline not just a new era of rapid innovation, but also growing fears that AI could ultimately shrink job opportunities in the tech industry.

“If you were starting whatever you’re starting 20 years ago, you would have had to have built up all these different competencies inside your company, and now there are just great platforms to do it,” Zuckerberg said.

He argued that AI is allowing smaller, more focused teams to achieve what once took entire departments.

“This is just going to lead to much better quality stuff that gets created around the world,” he said. “You’re just being able to have these, like, very small talent-dense teams that are, like, passionate about an idea.”

But while that message sounds like a celebration of innovation, it also carries a stark warning that the future of tech may be built with far fewer workers.

In a separate appearance on The Joe Rogan Experience podcast earlier this year, Zuckerberg predicted that by 2025, Meta and its competitors will deploy AI systems that can effectively perform the job of a “midlevel engineer” — a layer of the tech workforce that has historically formed the backbone of software development.

“Probably in 2025, we at Meta, as well as the other companies that are basically working on this, are going to have an AI that can effectively be a sort of midlevel engineer that you have at your company that can write code,” he said.

That forecast feeds directly into one of the biggest concerns sweeping the tech industry: that as AI becomes more powerful, it may displace millions of knowledge workers, beginning with those in software engineering, customer service, data entry, and IT operations.

Some researchers say that the future is already taking shape, but with some challenges. Harry Law, an AI researcher at the University of Cambridge, warned that large language models (LLMs) may appear capable but often produce buggy or insecure code.

“Ease of use is a double-edged sword,” Law told Business Insider. “Beginners can make fast progress, but it might prevent them from learning about system architecture or performance.”

He added that overreliance on AI for coding could make applications harder to scale and debug, while opening the door to security vulnerabilities.

However, major tech companies are racing to automate their software development pipelines. Google CEO Sundar Pichai disclosed in October that more than 25 percent of the company’s new code is now AI-generated and reviewed by engineers.

“This helps our engineers do more and move faster,” Pichai said during the company’s third-quarter earnings call. He called the shift a major boost to Google’s “productivity and efficiency.”

Others have gone even further. Shopify CEO Tobi Lütke reportedly instructed company managers to prove that AI couldn’t perform a task before seeking approval to hire a new employee. That policy, viewed by some as radical, reflects the growing push across Silicon Valley to lean into AI while reducing labor costs.

Zuckerberg himself declared 2023 the “year of efficiency” at Meta, a year marked by several waves of mass layoffs that saw thousands of workers lose their jobs. In that context, his praise for small teams and automation has drawn criticism from those who see AI as a justification for downsizing.

The startup world, however, is embracing the shift with enthusiasm. Y Combinator CEO Garry Tan said in March that startups are now reaching $10 million in annual revenue with teams as small as five to ten people.

“The wild thing is people are getting to a million dollars to $10 million a year revenue with under 10 people, and that’s really never happened before in early stage venture,” Tan said in an interview with CNBC.

He described the phenomenon as being powered by “vibe coding,” a term coined by OpenAI cofounder Andrej Karpathy. In a February post on X, Karpathy explained the term as a style of working where developers no longer write code in the traditional sense but instead guide AI models through prompts, copy-paste snippets, and let the machine handle the heavy lifting.

“It’s not really coding,” he wrote. “I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.”

Anthropic cofounder and CEO Dario Amodei has gone so far as to say AI could be “writing essentially all of the code” within 12 months.

OpenAI CEO Sam Altman echoed that sentiment in February, saying he expects software engineering to look “very different” by the end of 2025.

These declarations reflect a broader cultural shift in tech, where large companies are no longer prioritizing headcount as a measure of strength. Instead, they are optimizing for smaller, faster, more efficient teams — often powered by AI.

However, that evolution raises uncomfortable questions about the future of employment. If AI becomes proficient enough to replace mid-level engineers, what happens to entry-level developers? How will the next generation of programmers build expertise if they are never given the chance to work on real systems?

And beyond engineering, as AI tools become more capable in design, marketing, legal analysis, and business operations, will entire swaths of white-collar work be redefined, or eliminated?

In a settlement that underscores the growing scrutiny of Big Tech’s data practices, Google has agreed to pay $1.375 billion to the state of Texas to resolve claims that it violated residents’ privacy rights by unlawfully tracking and storing sensitive user data.

The announcement was made on Friday by Texas Attorney General Ken Paxton, who framed the deal as the largest privacy-related settlement ever secured by a state against the tech giant—and a defining moment in his office’s ongoing battle against what he described as Silicon Valley’s disregard for consumer rights.

“For years, Google secretly tracked people’s movements, private searches, and even their voiceprints and facial geometry through their products and services,” Paxton said in a statement. “This $1.375 billion settlement is a major win for Texans’ privacy and tells companies that they will pay for abusing our trust.”

The settlement stems from two separate lawsuits filed by Paxton’s office in 2022. The suits accused Google of systematically collecting personal data without adequate user consent, including biometric identifiers and location information, in violation of Texas’s biometric privacy laws and the state’s Deceptive Trade Practices Act.

Central to the claims were allegations that Google’s Chrome browser’s incognito mode misled users into thinking their activity was private, and that location data was being collected through Google Maps even when users had turned off location history settings. Additional claims were tied to the collection of biometric data through Google Photos, where users’ facial geometry and voiceprints were allegedly used for features like facial recognition without informed consent.

Though Google denies any wrongdoing and has admitted no liability as part of the settlement, the financial penalty is a sign that state attorneys general are willing to challenge powerful tech companies over how they collect and use consumer data.

Google spokesman José Castañeda said the settlement “resolves a raft of old claims” that concern product policies which have “long since changed.” He emphasized that Google will not be required to alter any of its current services as a result of the agreement.

“This settles a raft of old claims, many of which have already been resolved elsewhere, concerning product policies we have long since changed,” Castañeda said. “We are pleased to put them behind us, and we will continue to build robust privacy controls into our services.”

However, the settlement’s magnitude, nearly $1.4 billion, eclipses previous agreements reached between other states and Google. It follows a similar $1.4 billion settlement Paxton secured from Meta Platforms Inc. in 2023, involving the unauthorized use of facial recognition technology on Facebook and Instagram.

Texas Leading a New Front on Tech Regulation

The back-to-back billion-dollar deals position Texas as a leader in holding tech firms accountable for data privacy violations, particularly through enforcement of its state-level biometric privacy law. Paxton’s legal strategy has relied on leveraging Texas’s biometric statute, which is modeled in part after Illinois’s Biometric Information Privacy Act (BIPA)—a law that has already led to several high-profile legal outcomes.

“Big Tech is not above the law,” Paxton said.

Although federal legislation on data privacy has remained stalled in Congress, the Texas settlement demonstrates how states are stepping into the regulatory vacuum. Experts say the rising tide of state-led enforcement could create a patchwork of privacy regimes nationwide, and increase pressure on companies to build more transparent and accountable data systems.

A Warning to The Industry

The settlement sends a strong message to other tech companies relying on user data to power their services. As regulatory scrutiny increases at both state and federal levels, industry players are likely to face mounting legal costs and reputational risks if found to be in breach of local privacy laws.

What’s more, the deal reinforces the growing relevance of biometric privacy—a subset of data rights that encompasses facial recognition, fingerprint scanning, voice recognition, and more. With tech companies increasingly incorporating such features into their products, legal safeguards around biometric data may become the next major battleground.