The Ladder Is Evaporating

The career path that built the American middle class is about to disappear — not from the top, but from the bottom.

We tell young people the same story we were told: get the degree, land the entry-level job, work hard, learn the ropes, prove yourself. The ladder is right there, so just grab the first rung and start climbing.

That story is breaking apart. Not slowly, not theoretically, not in some distant future that gives us time to adjust. It’s breaking apart right now, and the people most affected, the twenty-somethings about to walk across a graduation stage, are the ones who will suffer the consequences.

I see this from two vantage points that shouldn’t conflict but increasingly do. As the CEO of B:Side Capital, I see the math. I see what AI agents can already do, what they’re about to do, and what that means for every hiring decision I’ll make going forward. As a professor at ASU, I see the faces. I see bright, hungry students who did everything right and are about to graduate into a world that may not have a rung for them to grab. The dissonance between these two views keeps me up at night, and I can’t stay quiet about it anymore.

What Just Happened

This week, Mustafa Suleyman, the CEO of Microsoft AI, co-founder of DeepMind, and one of the people closest to the actual technology, sat down with the Financial Times and said something that should stop you cold. He predicted that AI will achieve human-level performance on most white-collar professional tasks within twelve to eighteen months. Accounting, legal, marketing, project management, anything that involves sitting down at a computer. All of it, within eighteen months.

He’s not some LinkedIn influencer farming engagement. He’s the guy building the thing, and he’s far from alone. Anthropic CEO Dario Amodei warned last year that AI could eliminate half of all entry-level white-collar jobs. Ford CEO Jim Farley said it would cut the total number of white-collar positions in America by fifty percent. Matt Shumer, an AI researcher whose essay went viral this week, compared this moment to February 2020, right before the pandemic hit, except he thinks this will be more dramatic.

Now, I know what some of you are thinking. We’ve heard this before. The predictions are always too aggressive. The technology never quite delivers on the hype. Fair enough. Critics rightly point out that automating a task is not the same as automating a job. Studies have shown that AI tools still stumble on real-world office work, and one recent study found AI actually made software developers slower, with tasks taking twenty percent longer. Profit margins outside of Big Tech haven’t budged. We’re not living in the apocalypse yet.

But here’s the truth: the direction is right, even if the timeline is off by a year or two. If you’re making career decisions, hiring decisions, or investment decisions based on the assumption that this wave will politely wait for you to get ready, you are making a catastrophic mistake. The ladder that generations of Americans climbed is evaporating, not from the top, but from the bottom.

The View From Both Sides

Let me tell you what I see from each of my worlds, because the dissonance is deafening.

From the CEO’s chair, the math is becoming impossible to ignore. I can spin up an AI agent that works twenty-four hours a day, never calls in sick, never needs onboarding, and doesn’t have a learning curve that takes six months before it produces real value. It costs a fraction, maybe ten percent, of what I’d pay a junior hire. I don’t say this with glee. I say it because it’s true, and because every founder and executive I talk to is running the same calculation, even if they won’t say it out loud. The question that used to be “who should we hire?” is becoming “do we need to hire at all?” When a company can spin up a thousand agents in the time it takes to post a job listing, the economics of entry-level human employment start to break down.

From the professor’s chair, the view is more personal and more painful. I look at my students and I see real talent, people who did everything right, went to a good school, worked hard, learned how to think and build and lead. The brutal truth is that the world they prepared for may not exist by the time they walk across the stage, and no one is preparing them for that reality. We’re still running the old playbook: résumés, interview prep, networking events. We’re acting as if the job market they’re entering bears any resemblance to the one that existed five years ago. It doesn’t, and pretending otherwise is a disservice to every student sitting in every classroom in this country.

The Uncomfortable Question

So let’s say it plainly.

If you’re a hiring manager and you can get the output of three junior analysts from one AI system at a tenth of the cost, are you hiring those three analysts? Of course you’re not. If you’re a law firm partner and your AI can draft contracts, review documents, and conduct legal research faster than a first-year associate billing at four hundred dollars an hour, are you hiring that associate? Maybe one, but not ten. If you’re running a marketing department and an AI agent can produce campaign copy, analyze performance data, build audience segments, and generate creative variations in real time, do you need a team of eight, or do you need two people who know how to direct the machine?

These aren’t hypotheticals. These conversations are happening right now, in real boardrooms, at real companies. Every person in a hiring position at virtually every company (media, tech, finance, consulting) is asking the same question: Is this even a job anymore, or can we use AI for this?

The first rung is already gone, and Suleyman is saying the rest aren’t far behind.

I’m Not Here to Debate the Remedies

I want to be clear about what this piece is and what it isn’t. I’m not here to debate UBI, or to forecast the social chaos that may follow when a significant slice of the knowledge workforce gets displaced in a compressed timeframe, or to talk about the many ways AI could harm us beyond employment: the safety concerns, the concentration of power, the erosion of meaning. Those are real conversations, important ones, but they’re not this conversation.

What I want to do, and what I feel a responsibility to do given where I sit, is issue a direct, practical, no-nonsense call to action for two groups of people. Those of you who currently hold jobs that AI is coming for, and those of you, my students especially, who are about to step into a job market that is transforming beneath your feet. The obstacle right now is enormous, but the way through it is clearer than you might think.

If You’re a Student About to Graduate

Stop optimizing for “getting a job” and start optimizing for being impossible to replace.

The old playbook was designed for a world where the bottleneck was access to knowledge and the ability to process information. That bottleneck is dissolving in real time. The new scarce resources are judgment, taste, relationships, and the ability to operate in ambiguity where the stakes are real and the context is messy. AI can process information faster than any human who ever lived, but it cannot be accountable. It cannot sit across from a client who just lost a family member and figure out how that changes the estate plan. It cannot read the room.

Think about what that means practically. AI is extraordinary at tasks that live entirely on a screen, but it is terrible (and will be for a long time) at anything requiring physical presence, human trust, or navigating the chaos of the real world. A college student who combines a white-collar education with the ability to operate in the physical world has an edge that is growing, not shrinking. Lean into the physical-digital gap. The students who will thrive aren’t the ones who can do what AI does, but the ones who can take AI’s output and do something with it: evaluate whether the legal brief is actually right, tell a client why this marketing strategy matters for their specific situation, spot when the financial model has a garbage assumption buried in row forty-seven. The skill isn’t doing the work. It’s knowing what good work looks like and being accountable for the outcome.

Here’s the irony that should give you hope: AI makes it easier than ever for a solo operator or a tiny team to punch way above their weight. If traditional employment is becoming less reliable, the ability to create value independently becomes a form of economic insurance. One person with the right tools can now do what used to take a team of ten, which means the entrepreneurial path has never been more accessible or more necessary. Build something, and build your network before you need it. The people who weather economic dislocations are the ones deeply embedded in communities of people who trust them and want to work with them. Your network is your safety net, not your résumé. That’s unsexy advice, but it is the most durable advice I can give you.

If You’re Already in The Workforce

Here’s a shift most people miss: if you currently have a job, that job is your single greatest asset. Not because of the paycheck, though that matters, but because of what it gives you. It gives you a funded laboratory. You have a stable platform, a paycheck, health insurance, and most importantly, a live environment in which to learn the most consequential skills of your professional life. Your number one priority right now is to keep that job, not out of fear, but out of strategy.

Think about it: every company in America is about to go through an AI transformation, whether they know it yet or not. Most organizations are in a weird limbo right now where leadership knows AI is coming but rank-and-file adoption is patchy and uneven. That limbo is your window. The person who figures out how to use these tools to genuinely ten-x their output doesn’t get automated; they become the person who leads the automation. You want to be that person, so far ahead of the curve that when leadership looks around and asks “who understands this stuff?”, your name is the first one that comes up. Don’t wait for your company to train you and don’t wait for permission. Start now, document the results, and become the AI power user in your department before someone else does. You won’t just keep your job. You’ll become essential.

But keeping your job is just step one. The real goal is to become so valuable that your company can’t imagine operating without you. The deeper move here is about identity. If you think of yourself as “I write contracts” or “I build financial models” or “I create marketing copy,” you’ve defined yourself by a task, and tasks get automated. If instead you think of yourself as “I help companies manage risk in complex deals” or “I help leadership make capital allocation decisions under uncertainty,” you’ve defined yourself by an outcome. Outcomes require judgment, context, stakeholder management, and accountability. AI can do tasks, but it can’t own outcomes. That distinction is the difference between being replaceable and being the person your organization builds around.

That’s the standard you should be aiming for. Not “good enough to keep my seat,” but the kind of person who combines deep domain expertise with AI fluency and the judgment to know when the machine is right and when it’s dangerously wrong. The kind of person who can walk into any room and create value that didn’t exist before they showed up. The further you are from revenue, from the client relationship, from the sale, from the decision that generates revenue, the more exposed you are. Back-office analytical work is the first to go, while client-facing, relationship-driven, revenue-generating work is the most protected. If your current role is purely internal and analytical, find ways to get in front of clients. Even a lateral move that puts you closer to the customer can change your trajectory overnight.

Here’s where this all comes together: the skills that make you indispensable inside your company are the same skills that make you indispensable everywhere. AI mastery, domain expertise, client relationships, the ability to own outcomes, these aren’t just survival tools. They’re the building blocks of an elite career, whether that means rising within your current organization, leading a new division, or someday building something of your own. The person who can combine AI tools with human judgment and accountability doesn’t just survive the transformation; they become the person their company turns to when the stakes are highest. Your current job is where you develop that combination, so treat it that way.

Here’s the part nobody wants to hear: some roles really are going away, and no amount of upskilling will preserve them in their current form. If the work essentially amounts to “take information from System A, apply a set of known rules, and produce output in System B,” that function will not exist in three to five years regardless of what anyone does. The mature response isn’t denial. It’s using the opportunity you have right now to evolve into someone who does the work that can’t be automated: leading teams through ambiguity, managing client relationships, making high-stakes decisions where the data is incomplete and the consequences are real.

The Bottom Line

Here’s what I know from standing in both worlds.

The worst thing you can do is pretend this isn’t happening, and the second worst thing is to panic and assume all is lost. The people who thrive in disruption are the ones who move toward the change rather than away from it. The instinct is to protect what you have, but the better instinct is to position yourself where the value is moving to. The obstacle in front of you right now, this technological transformation rewriting the rules of work, value, and human contribution faster than any of us expected, isn’t blocking your path. It is your path.

I’ll be writing much more about this in the weeks ahead: what the new career architecture actually looks like, how the relationship between humans and organizations is going to restructure, and what the emerging playbook is for building economic resilience when traditional employment may no longer be the default.

But the first step is the simplest and the hardest: stop pretending the ladder is still there. It’s not, and the question is what you’ll build in its place before you have to.