The AI Gold Rush Is Eating Silicon Valley Alive

Oracle just fired thousands of people. CEOs are suddenly discovering AI makes great cover for mass layoffs. OpenAI raised another $12 billion at a $730 billion valuation that makes my calculator weep.

Welcome to 2025, where artificial intelligence has become Silicon Valley’s most expensive midlife crisis.

The money flowing into AI right now defies rational explanation. OpenAI’s latest funding round brings their total raised to $122 billion — for a company that burns cash faster than a Tesla in a Norwegian winter. That $730 billion valuation puts them ahead of Tesla, ahead of Meta, ahead of companies that actually, you know, make consistent profits.

But here’s what’s really happening: We’re watching the most dramatic reshuffling of tech priorities since the iPhone launch in 2007. Companies are cutting humans to fund AI experiments. States are racing to regulate something most politicians barely understand. And somewhere in China, 100 robotaxis simultaneously lost their minds and gridlocked traffic.

The parallels to the dot-com bubble are impossible to ignore, except this time the bubble is filled with neural networks and powered by FOMO.

The Great AI Job Massacre

Close-up of a chocolate dessert bowl with slices of chocolate bar, fruits, and cream. Photo by thecactusena ‎ / Pexels

Oracle’s mass layoffs tell a story Silicon Valley doesn’t want to admit: AI adoption means human subtraction. The company hasn’t responded to requests for comment about exactly how many thousands got axed, which in PR terms usually means “more than you think.”

What’s new isn’t the layoffs themselves. Tech companies have been cutting jobs for two years as high interest rates made cheap money extinct. What’s new is the messaging.

More CEOs are explicitly blaming AI for job cuts, according to recent reporting. Not “economic conditions” or “market headwinds” — AI. As in, “We’re replacing you with software, and we’re not even pretending otherwise anymore.”

This represents a seismic shift in how tech companies talk about automation. For decades, the party line was that technology creates more jobs than it destroys. Remember all those think pieces about how ATMs actually increased bank employment? Those arguments haven’t disappeared, but they’ve gone quiet.

I think we’re seeing the end of tech’s jobs theater. The charade where companies pretended AI would “augment” human workers rather than replace them. Oracle’s cuts suggest we’ve moved past augmentation into straight replacement territory.

The timing isn’t coincidental. Companies need massive capital to compete in AI, and labor costs are the biggest line item on most balance sheets. Fire 10,000 people making $150,000 each, and you’ve freed up $1.5 billion annually for GPU clusters and machine learning talent.

It’s brutal math, but it’s honest math.

When Robotaxis Attack

Meanwhile in China, Baidu learned why “move fast and break things” hits different when your things are autonomous vehicles in traffic.

At least 100 of Baidu’s robotaxis malfunctioned simultaneously, creating what I imagine was either the world’s most expensive traffic jam or performance art about the fragility of automated systems. Baidu hasn’t commented, which tells you everything about how well this went.

Autonomous vehicles represent AI’s highest-stakes testing ground. Get computer vision wrong on a chatbot, and someone gets bad restaurant recommendations. Get it wrong on a robotaxi, and people die.

The mass malfunction reveals something tech companies hate acknowledging: AI systems fail in correlated ways. Human taxi drivers don’t all stop working at once because they use different brains. But when your entire fleet runs on the same software stack, a single bug can brick your entire operation.

This is the nightmare scenario for any company betting on AI infrastructure. All the redundancy and fail-safes in the world can’t protect you from systemic software failures.

My read: We’re about to see a lot more of these coordinated AI failures as deployment scales up. The question isn’t whether they’ll happen, but whether companies can contain the damage when they do.

Businessman reading a financial newspaper at a desk, highlighting finance and commerce theme. Photo by nappy / Pexels

The Regulation Stampede

States aren’t waiting for federal leadership on AI regulation. Despite Trump ordering them to back off, jurisdictions from California to Utah are plowing ahead with their own AI guardrails.

This creates a fascinating dynamic. Trump wants to position America as the global leader in AI development, which typically means fewer regulations. But states are looking at Oracle’s layoffs and Baidu’s traffic jams and thinking maybe some rules might be helpful.

California has been particularly aggressive, which makes sense given how many AI companies call it home. When your backyard is ground zero for an emerging technology, you get to see both the promise and the carnage up close.

The state-federal tension here mirrors cannabis legalization from the 2010s. Federal prohibition, state experimentation, eventual federal capitulation. I’d bet we see a similar pattern with AI regulation, where states become the testing ground for policies that eventually go national.

What’s different is the timeline. Cannabis policy evolved over decades. AI regulation is happening in months. The technology is moving too fast for the usual legislative pace.

While everyone obsesses over AI, social media companies are dealing with their own existential crisis. Australia’s under-16 ban is forcing platforms to actually enforce age restrictions, and early results suggest they’re not great at it.

Facebook, Instagram, Snapchat, TikTok, and YouTube are all struggling to comply with the new rules, according to Australia’s eSafety regulator. This shouldn’t surprise anyone who’s ever seen a 12-year-old with a TikTok account, but it highlights how much social media business models depend on young users.

The Australia ban is a test case for similar legislation worldwide. If platforms can’t figure out age verification there, they’ll face the same problems everywhere else considering similar rules.

Instagram’s recent retreat from “PG-13” branding shows how seriously they’re taking regulatory pressure. When the Motion Picture Association pushes back on your marketing language, you know you’ve wandered into territory lawyers care about.

These regulatory pressures are creating interesting dynamics. Social media companies need AI to handle content moderation at scale, but AI regulation threatens to limit how they can use those tools. Meanwhile, they’re cutting costs to fund AI investments while dealing with shrinking user bases due to age restrictions.

It’s like trying to solve a Rubik’s cube while someone keeps changing the colors.

The Global Internet Fracture

Detailed close-up of a newspaper displaying global financial market statistics and country flags. Photo by Markus Spiske / Pexels

Russia’s internet restrictions show how AI and geopolitics are reshaping the global web. As the Kremlin spends heavily on censorship technology, Russians are scrambling for new circumvention tools.

This cat-and-mouse game between authoritarian governments and internet users has been going on for years, but AI is changing the rules. Machine learning makes censorship more sophisticated and harder to evade. But it also makes circumvention tools smarter.

The result is an arms race between AI-powered censorship and AI-powered freedom tools. Whoever has better algorithms wins, at least until the other side catches up.

This dynamic extends beyond Russia. China’s Great Firewall uses machine learning. Iran’s internet restrictions rely on AI pattern recognition. Even democratic countries are deploying AI for content moderation that looks a lot like censorship depending on your perspective.

We’re moving toward a world where internet access depends on your government’s AI capabilities. That’s either fascinating or terrifying, depending on where you live.

The Money Problem

OpenAI’s $12 billion funding round reveals the central tension in today’s AI boom: Nobody knows how to make money from this stuff.

$730 billion valuations don’t happen because investors expect steady dividend payments. They happen because people believe AI will fundamentally restructure the global economy, and they want ownership stakes before that happens.

But here’s the uncomfortable truth: Most AI applications today are cost centers, not profit centers. ChatGPT costs OpenAI millions to run. Autonomous vehicles require massive infrastructure investments. Computer vision systems need expensive hardware and constant updating.

The bet is that AI will eventually generate enough value to justify these investments. Maybe through subscription services, maybe through productivity gains, maybe through entirely new business models nobody’s invented yet.

I think this bet will eventually pay off, but not for everyone placing it. The current funding environment reminds me of cleantech investing in the late 2000s — lots of capital chasing breakthrough technologies that mostly broke wallets instead of markets.

The difference is scale. Cleantech was a sector. AI is supposed to be everything.

What Happens Next

We’re about six months into what I’d call the “AI reality phase” — the period where deployment meets actual user needs rather than investor fantasies.

Oracle’s layoffs suggest companies are serious about restructuring around AI, not just experimenting with it. The robotaxi malfunction shows we’re pushing AI into high-stakes environments before the technology is ready. State regulation efforts indicate governments are done pretending AI governance can wait.

The next 18 months will determine whether we’re in a genuine technological transition or just an extremely expensive bubble. The difference matters for everyone from Oracle’s laid-off workers to the investors funding OpenAI’s burn rate.

My prediction: We’ll see more coordinated AI failures like Baidu’s robotaxi mess. More companies will use AI as cover for mass layoffs. More states will ignore federal AI policy preferences. And more money will flow into AI startups that have no clear path to profitability.

The question isn’t whether AI will transform the economy. It will. The question is whether the current approach — throwing infinite money at the problem while cutting human jobs — actually gets us there faster.

Kris Jenner and the Attention Economy

In the middle of all this technological upheaval, hundreds of thousands of Chinese social media users are sharing images of Kris Jenner for good luck.

This feels like a perfect metaphor for our current moment. While engineers debate neural network architectures and investors throw billions at AI startups, regular people are just trying to figure out how to navigate an increasingly weird world.

The Kris Jenner phenomenon shows how global culture flows through social media in ways that have nothing to do with Silicon Valley’s grand AI ambitions. People want community, meaning, and hope for prosperity. Sometimes that means sophisticated AI tools. Sometimes it means sharing pictures of reality TV stars.

The gap between what tech leaders think people want and what people actually want has never been wider. That gap represents either a massive market opportunity or a fundamental misunderstanding of human nature.

I’m betting on misunderstanding.

Looking Back, Moving Forward

Chris Espinosa’s 50-year tenure at Apple provides useful perspective on Silicon Valley’s current AI obsession. In 1976, he was 14 years old, riding a moped to demonstrate computers made in Steve Jobs’s garage. Apple’s market cap was effectively zero.

Fifty years later, Apple is worth over $3 trillion. The computers Espinosa demonstrated have become iPhones carried by billions of people. The company has changed “a bit,” as the headline dryly notes.

The lesson isn’t that current AI investments will inevitably pay off like Apple did. Most companies from 1976 are dead. The lesson is that transformative technology takes decades to mature, and the biggest winners often aren’t obvious during the hype phase.

Apple survived and thrived because it focused on building products people actually wanted to use, not just technology that impressed other engineers. The companies that survive the current AI boom will be the ones that solve real problems for real people, not the ones with the highest valuations or the most impressive demos.

That focus on human needs feels absent from much of today’s AI development. Oracle fires thousands of people to fund AI initiatives. OpenAI raises billions for technology that costs more to run than it generates in revenue. Robotaxis malfunction in traffic because nobody asked whether the technology was ready for deployment.

What I’m Watching

Oracle’s Q3 earnings in March: Will they quantify exactly how many jobs AI eliminated, and will the cost savings translate to AI revenue? If they can’t show positive ROI on their human-to-AI swap, other companies might pause their own replacement strategies.
Baidu’s next autonomous vehicle deployment: How they handle the mass malfunction fallback will signal whether robotaxi companies can manage systematic failures or if they’re too systemically fragile for widespread adoption.
California’s AI regulation timeline through summer 2025: Whether state-level rules can actually constrain AI development or if companies just move operations to more permissive jurisdictions. The enforcement mechanisms matter more than the rules themselves.
OpenAI’s revenue disclosure by year-end: At a $730 billion valuation, they need to show a path to profits that doesn’t require burning $12 billion every funding cycle. The gap between valuation and revenue has to close eventually, and when it does, it’ll reshape the entire AI investment landscape.