A War On Talent

What was my story of the year? I’m a tech guy by nature, it’s just who I am. And one defined 2025, the cover of Time says it all. Look how far we’ve come.

Years back, we lived in a very different world, and I worked at a remarkable and somewhat unconventional place. Interacting with clients. Solving problems. In meetings, I’d often use a pencil with a worn eraser to jot down these quaint notes. No Field Notes, no Moleskine journal, just a spiral-bound notebook. How times have changed. Consider this:

• Video calls weren’t transcribed; hence the notebooks. Granted, I still take notes because I’m the type of guy who believes the act of writing helps me remember instead of an AI transcribing what I should remember. Call me old school.
• Sales calls were an in-person event, often leadership traveled hundreds of miles to establish long-term relationships.
• Return to Office (RTO) wasn’t a mandate anyone thought about. We went. Or didn’t. Technically, we took pride in being big boys and girls, driving deadlines and excellence. Anything to hit a goal.

Then March 202 hit, and nobody left the house. Remote work rewired the game. Workers realized they had true options.

These were the dark days; we all did the best we could in quiet solitude. About halfway through this era, I showed up for work, hopped on a series of video calls, and three high-caliber performers, one after another, asked, rather demanded, for raises. I’m not saying they didn’t deserve pay bumps. Many did; a hot technology company in a unique space. However, I grew up in an age where performance reviews occurred on a timeline, we calibrated against peers for fairness, and managers often fought over limited funding pools.

A certain process existed.

But that’s not this story. Solid contributors wanted thirty-point bumps, or they had places to go. The market was on chaotic fire.

I’d like to believe management and human resources worked quickly, doing everything to stay ahead of the curve. Some stayed. Others moved on. The lesson here is all employers need to truly analyze their arrangement. It’s more than money. Personally, I’ve always had the tagline that employment is a two-way street. Yes, employers have power in the relationship. But often that is superficial, or it was. You see, the employee chooses who they work for and that choice varies in degrees, somewhat like a statistical confidence interval. That’s why tech companies once bent over backwards advertising perks, flexible schedules, and advanced compensation packages. If these fell outside the confidence window against their peer groups, well, migration happens leading to competitive disadvantages. Numerous cases, and accusations, of employer collusion exist. Apple. Google. Microsoft. That’s why the age of perks sprung from the ether, granted there were also certain strings attached—don’t you dare believe otherwise.

The Thinking LLM

That equilibrium held for a short while, but it was just the opening act. The real disruption was loading or tuning.

Then the machines arrived.

And the rules short-cicruited.

This is not a Matrix-style invasion. But these agents are blurring traditional lines of employment. AI tooling has democratized technical work. Now, business teams can spin up servers, run analyses, build prototypes—work that once required dedicated, and sometimes lacking, dedicated IT resources. And vice versa. Frankly, it’s easier to switch hats while having the machines do the heavy lifting.

This is more than a wholesale flattening. And, in many cases, depending on the line of work, these processes and changes are gutting roles. Mostly, it’s impacting once sacred cows: the younger generation. If you don’t believe me, the unemployment statistics are unprecedented since ChatGPT burst onto the scene. Ask any parent with kids in college.

Some will argue that entry-level roles aren’t needed, that’s what a vibrant market does; it adjusts and weeds out weaknesses. But I’d counter we’re not thinking through the long game. How is a young person supposed to become a star sales leader if they aren’t given a chance, a tough territory? What about the project manager? Or junior consultant? Yeah, certain entry-level work matters because it compounds. Have we thought about what happens when the pipeline of young workers doesn’t exist? Where do the leaders of tomorrow come from?

I suppose we’ll just fail fast and see what happens later, ignoring all risks.

It won’t be the first time. Or the last time. The United States economy is a capitalistic petri dish and this is what it does, with limited government guidance.

Big Tech Debt Cometh

The human cost is one side. The balance sheet tells the other.

Despite the pessimism, I love AI’s promise. Heck, it powers this site; the advanced author analytics come from machine learning algorithms. Yes, the world moving at breakneck speed is frightening. I agree.

Yet, over the past few weeks, the far underreported story in tech hasn’t been a flashy product release or new AI model. It’s been a borrowing spree.

Amazon, Meta, Google, Oracle, and others have collectively raised nearly $100 billion in new debt, which is an aggressive move for companies sitting on piles of cash. The stated purpose is relatively straightforward: fund the next era of AI infrastructure, from data centers to GPU clusters to the industrial-scale computing required for model training.

Maybe this is nothing but the cost of doing business; earnings are through the roof and stocks are at an all-time high.

But what’s a bit different is that these companies are spending faster than their operating cash on hand supports. Rather than upset shareholders with dilution or shrinking margins, they’re borrowing. This is unusual for cash-rich companies.

Why does this matter?

It signals a shift from AI hype to AI spending reality. Building the hardware footprint for frontier models requires staggering capital, and the hyperscalers are scrambling to secure capacity and make agreements before competitors do. That urgency explains the debt. But it also reveals a hard-to-ignore tension. Yes, AI adoption is real, but the money coming in lags far behind the money going out. When firms this large take on leverage, it usually means they’ve hit a wall. Something has to give.

And that’s where the human cost comes in.

When spending spikes, companies cut people. Headcount becomes the release valve. Every one of the big issuers has already gone through rounds of layoffs over the past 18 months.¹ Yet this borrowing spree suggests more are coming and WARN notices, legally required layoff alerts in certain states, have already been released. Shareholders expect margin protection. Boards demand efficiency. AI promises automation. Roles that once seemed safe suddenly feel like liabilities.

This debt explains the urgency. The layoffs explain the cost.

Yet, whatever happened to AI finding a cure for cancer? Or building a fusion reactor?

This isn’t just an investment story, it’s a restructuring odyssey. The AI arms race is forcing the industry to scale hardware while shrinking payrolls, accelerating a shift that will reshape the tech labor market throughout 2025 and 2026. If Big Tech can’t make AI work, who can? The better question: If Big Tech can’t streamline operations with this stuff, why should Procter and Gamble or John Deere try?

A War on Talent

Layoffs haven’t slowed because the underlying pressures that triggered them never actually went away. Companies cut heavily in 2022 and 2023, but those rounds removed redundancies: overlapping teams, pandemic-era overhiring, and experimental projects that never found traction. That darned Metaverse never panned out despite billions being pushed into it. Although, we did get awesome Ray-Ban glasses out of the deal.

What’s happening now feels deeper. Firms are restructuring their entire operating models around efficiency, automation, and a smaller, more senior workforce. Middle management is getting thinned because of sink theory. Legacy product groups are winding down. Roles far from core revenue are being deemphasized. This isn’t a cyclical correction; rather, it’s the industry resizing itself for a new era. Isn’t this what the people wanted?²

Another driver is organizational maturity. The big platforms spent a decade expanding aggressively into every adjacent market they could touch. Healthcare. Insurance. Manufacturing. Now they’re reversing direction. Companies are shutting down slow-growing divisions, consolidating product lines, and centralizing engineering instead of scattering small teams across organizational charts. Centralization and decentralization is a thing, a seesaw for change. As internal structures tighten, headcount drops. And unlike past years, when layoffs were often followed by new waves of hiring, companies no longer feel the need to replenish those roles. The expectation? Modern workflows will handle the load, until they don’t.

The other reality is cultural; remember Elon and his sink. Wall Street rewards leanness over ambition. Even when profits are strong, companies are being pushed toward smaller, sharper organizations with clearer financial discipline. Executives have learned that layoffs reliably lift stock prices, at least in the short term, and that creates an incentive to keep on trimming, evolving, or turning headcount into a standing target rather than a sacred asset. In the words of Master Yoda, Begun the war on talent has.

The Bailout Cometh

None of this is new. It’s the scale that’s different.

This is a cycle. It’s happened before; heck, more than half of the companies I’ve called home didn’t exist when I was in college.³

Nothing new here. This is America, baby.

It’s not unheard of for companies to take substantial business risk. Jeff Bezos loudly went on Jay Leno saying he makes zero money and that turned out well for him. He was playing a different game where his equity allows him to live large and pay little taxes. Amazon famously lost money for years building an infrastructure, leveraging tax incentives and questionable policy structures. Looking back, do we honestly feel the company paid enough in taxes to compensate for the tragedy of the commons—our cracking roads and byways? This is a fair question to ask ourselves. But I can get a package on my doorstep the same day if I click buy on my phone. That’s a technological marvel. Yes, there is precedent, for large tech companies to fund losing businesses. And then, the model flips toward profitability.

AI has this potential despite being little more than extremely costly SQL queries with unproven results.

Yes, OpenAI bleeds money, and it will continue to do so at an alarming rate until the true cost of the API call drastically decreases. It’s just math. The data centers. Copper wiring. The water. The power. Reportedly, the company made 15 billion last year, which is incredible, while losing nearly 100 billion.

What happens if an AI darling fails? It will happen. But when it does, who pays? The taxpayer? Debt markets? Or the institutional investor?

And with most stock market gains coming from the top ten companies, maybe, it’s all of us. Is that a fair trade?

This is a time where innovation and ingenuity cast a wonderful promise, one that may not pan out, but where standing in place is its own odd, nervous self-fulfilling prophecy. For people who thrive with entrepreneurship, change, and constant motion it’s a rare window, an opportunity for a new life. Doers do.⁴

One way or another, it’s going to come at us fast.

References, Mentions, and Other Stuff

A busy year:

A 2024 Review

Last year, my article of the year fell along similar lines. Here, I pontificated on about the rise of open source LLMs from China, specifically Qwen 2.5. A few months later DeepSeek launched, crashing the stock market briefly due to their high-efficiency. The Chinese did the incredible with input tokenization, which one would only do because of chipset limitations and trade embargo. Because of these models’ permissive licensing, I believe they are more widespread in US companies than most believe. And with our economy having so much invested in the fate of the Magnificent Seven or Eight (tech companies with few regulations), the Chinese will continue to flood the market with these throughout 2026. That’s what the country does best, steal Intellectual Property and grind it down to a lower price point. What I didn’t expect were the advancements by Anthropic and OpenAI to offset these gains. Sonnet, Opus, and GPT perform. It’s amazing how far along they’ve come.

Second Place Stories (What I Read, News Edition)

• Iran’s proxy strategy hit a wall. Hamas, Hezbollah, and the rest; well, they all faced Israel’s defenses and lost. The conflict revealed what many already knew: years of preparation and intelligence work matter more than grandstanding. Alas, the cycle continues.
• A wonky economy. They tell you tariffs will fill the treasury and bring the work back to town, but here’s what actually happens when the money comes in. The government spends it faster than a high school kid spends his first pay check, and the jobs that were supposed to return just stay gone. That’s the way these things usually work. But I didn’t write this story yet because of differing opinions, well, there’s not enough data, and, by the time we know, the mid-terms will be upon us. A wait and see, an analytical review project I’ll take on next year.
• Remember those astronauts? No, I didn’t forget about them. The ship had its problems, the team couldn’t come home, and the people on the ground were working day and night to figure out how to return everyone home safely. But here’s the thing about people who know their work; they somehow find a way. Space is cool again, the race is heating up both foreign and domestic.
• Waymo is coming. The story won’t be driverless cars (they are amazing), it will be how the tech companies tweak the risk-profile between safety and profitability.

Product of the Year (What I Tinkered With)

• Cursor changed the way people work with code, and that’s saying something when every tool promises to change everything. Now we’ll see what happens when the machines write everything for us.
• Claude Code. Here’s something you don’t hear every day: the command line is cool again, and the old ways come back because they still work. It’s like vinyl. And old 1935 Underwood Typewriters that keep going. Funny how that works.
• Boring Consumer Product Award Winners, The Nintendo Switch 2 and iPhone 17 Pro. Both are incredible, the latest pro doesn’t burn a hole in my pocket. As for the Switch, it can play games made three years ago flawlessly. Onward.

Movie (What I Watched)

• Worst marketing of the year belongs to the AMC Movie Pass. One Imax Movie each month? A nearly unlimited pass costs the same-it’s better than Netflix. So I’ve seen dozens of movies the last quarter. Wicked. Fantastic Four. Eternity. Good Fortune. The Roses. F1. The Running Man. All great. But if I had to recommend one, it would be Hamnet because… well… it’s a fictional depiction of the Bard. And I cried like a wee baby, needed a second box of Kleenexes. Maybe see this one at home instead of inside a full theatre, unless you’re completely comfortable with those rolling tears.
• My runners up would be Eternity, followed by Eleanor the Great. Sinners and The Life of Chuck are also fantastic. Caveat, I haven’t caught Marty Supreme. Oh, Timothee Chalamet with the long, flowing hair… you can still pull off the upset. A good year for the Big Screen.

Game (What I Played)

• Contra. What’s better? Nothing.
• If a game that launched in 1987 doesn’t qualify, I’d lean toward Hades II. Pick up and play, if you dare. The fight for Olympus will hook you. Note, like my book reviews, I have a large backlog and haven’t dug into Silksong. A long wait for Team Cherry, and I’ve decided to savor this one.

And Best Invention (What I Read)

• Gadgets and Cars, they didn’t delivery, not in 2025. But those things called books, they never fail me. Staring over my finished stack, I ran back the clock, rereading and finishing numerous fantasy epics. Tolkien. Brooks. Jordan. But there is nothing like The Devils by Joe Abercrombie. The hard cover version is perfectly put together—interior illustrations that advance the story, weighty paper (this is the good stuff, not whatever Dan Brown put on shelves). This beast was printed with care. And as for the story, there be werewolves. And they be violent. Extremely. A gory delight; nobody writes like this.

Footnotes