Tech Is Firing People and Hiring Machines: How to Invest in the AI Arms Race Without Picking a Winner

Something strange is happening in tech right now. Companies are laying people off at an accelerating rate while simultaneously pouring record amounts of money into artificial intelligence. Prediction markets put the probability of tech layoffs exceeding 494,000 this year at 87.1%. At the same time, the race to build the best AI model is more intense than ever, with betting markets giving Anthropic a 50% chance of having the leading AI by December 2026, Google at 28%, and OpenAI at just 10%.

This isn't a contradiction. It's a pattern economists call creative destruction, where an industry tears down old structures to build new ones. Companies are slashing non-AI headcount to fund the AI arms race. The money that used to pay software engineers maintaining legacy products is now buying GPU clusters and training data.

And several other prediction market signals confirm that the broader tech ecosystem is tightening up. SpaceX has only a 5% chance of IPOing by May and 25.5% by June, suggesting private capital markets aren't exactly throwing money around. Tesla's Optimus humanoid robot has just an 18% chance of going on sale by year-end. The Musk vs. Altman lawsuit sits at 37% probability of resolving in Musk's favor. The picture that emerges is a tech sector concentrating its bets, not spreading them.

The Creative Destruction Cycle

This pattern creates a self-reinforcing loop worth understanding:

1. AI models get more capable, creating pressure on every tech company to invest or fall behind.
2. Companies cut non-AI staff and redirect that budget toward AI compute and talent.
3. Those layoffs show up in the 87.1% layoff acceleration figure, which sounds alarming in isolation.
4. The redirected money flows into GPU purchases, cloud computing contracts, data center buildouts, and electricity.
5. That infrastructure spending makes AI models even more capable, restarting the cycle at step one.

The companies doing the firing are also the ones doing the buying. That means tech stock indices can look deceptively calm on the surface, because mega-cap gains from AI spending offset the damage from broader workforce contraction. Under the hood, there's enormous dispersion, meaning some companies are thriving while others in the same sector are quietly deteriorating.

The Shovels Strategy: Who Wins No Matter What

During the California Gold Rush, most individual miners went broke. The people who reliably made money were the ones selling pickaxes, shovels, and denim jeans to every miner regardless of whether they struck gold. The AI race has the same structure. Whether Anthropic's 50% probability holds up, or Google's 28% shot comes through, or OpenAI stages a comeback from 10%, they all need the same underlying infrastructure.

That infrastructure stack runs deep, from the machines that make the chip-making machines all the way down to the electricity that powers the data centers.

The chip monopoly. ASML is the sole manufacturer of EUV lithography machines, the equipment needed to produce advanced AI chips. TSMC, Samsung, and Intel all need ASML's machines. There is literally no competitor. Prediction market confidence: BUY at 75%. Risks include geopolitical friction around China export controls, the cyclical nature of semiconductor orders, and a valuation that already reflects monopoly pricing power.

The GPU kingpin. NVDA supplies the GPUs that every major AI lab uses for training and running models. Anthropic, OpenAI, Google DeepMind, xAI (which sits at 9.6% in best-AI markets) — they all train on NVIDIA silicon. The creative destruction cycle is directly bullish here: companies cutting human workers and replacing them with AI compute means more GPU purchases. BUY at 78% confidence. The main risk is valuation. At roughly 30x forward revenue, NVIDIA is priced for perfection, and if AI investment proves to be a bubble, this stock faces the most severe compression of any name on the list. AMD and custom chips from Google (TPUs) and Amazon (Trainium) are also gradually eroding NVIDIA's moat.

The custom silicon and networking layer. AVGO (Broadcom) designs custom AI accelerators for Google and Meta, plus dominates the networking chips that connect GPUs inside AI clusters. BUY at 80% confidence, the highest conviction infrastructure call. Risks center on customer concentration and VMware integration debt. ANET (Arista Networks) makes the high-speed switches that handle traffic between GPUs during training. BUY at 76% confidence, with the caveat that hyperscaler spending can be lumpy quarter to quarter.

The fab equipment makers. AMAT (Applied Materials) at 74% confidence and LRCX (Lam Research) at 71% confidence sell the deposition, etching, and inspection equipment that TSMC and Samsung use to manufacture AI chips. Lam has outsized exposure to HBM (High Bandwidth Memory), the memory stacks sitting on every NVIDIA GPU and currently a key bottleneck in AI chip production. Both face China export control risk and the inherent cyclicality of semiconductor equipment orders.

The power and cooling backbone. This is the layer most people forget about. AI data centers consume staggering amounts of electricity, and every training cluster needs industrial cooling. VRT (Vertiv) provides power distribution, thermal management, and UPS systems for data centers. BUY at 77% confidence. VST (Vistra) operates natural gas, nuclear, and solar generation that feeds directly into hyperscaler data centers. BUY at 70% confidence. CEG (Constellation Energy), the largest US nuclear operator, has signed landmark data center power agreements including Microsoft's Three Mile Island restart deal. BUY at 69% confidence. Nuclear power is uniquely attractive for AI data centers because it provides 24/7 carbon-free baseload generation. EATON rounds out the power infrastructure with switchgear and transformers, though its diversified business dilutes the AI thesis. WEAK BUY at 70% confidence.

The AI Competitors Themselves

GOOGL at 28% probability for best AI represents compelling value. Google has DeepMind, massive compute infrastructure, proprietary TPUs, and the distribution moat of Search, Android, and Cloud. They benefit from the AI arms race as both a participant and an infrastructure provider through Google Cloud. BUY at 72% confidence. The biggest risk is antitrust rulings that could force structural changes, plus the paradox that AI-powered search could cannibalize Google's own advertising revenue model.

AMZN operates AWS, the backbone cloud provider for major AI labs. Amazon invested over $4 billion in Anthropic, the current prediction market frontrunner. BUY at 72% confidence. AWS collects what amounts to infrastructure rent: every AI model trained or run on AWS generates revenue for Amazon regardless of which company's AI ends up winning. The 87% tech layoff acceleration actually helps Amazon's cost structure as talent costs normalize, and enterprise cloud migration accelerates when companies cut on-premises IT staff. Risks include aggressive competition from Azure and Google Cloud compressing margins, and the reality that Amazon's own AI models (Titan) haven't gained traction.

MSFT presents the most complicated picture. Microsoft backed OpenAI with $13 billion-plus, and Azure is OpenAI's cloud host. But prediction markets show OpenAI declining sharply, dropping 3.5% in 24 hours to just 10% for best AI. If Anthropic truly displaces OpenAI, Microsoft's investment faces impairment risk. And Anthropic is AWS-first, not Azure-first, meaning Microsoft loses if Anthropic wins. WEAK BUY at 62% confidence, largely on the strength of Azure's broader cloud business.

For data center real estate, EQIX (Equinix) is the world's largest colocation operator. WEAK BUY at 63% confidence, limited by interest rate sensitivity (it's structured as a REIT) and the trend of hyperscalers building their own campuses.

One notable absence: Super Micro Computer (intended ticker SMCI, listed as WDFC in the data) assembles AI server racks and sits at the right place in the value chain, but accounting irregularities, auditor changes, and delayed SEC filings make it uninvestable until governance issues resolve. NEUTRAL at 45% confidence. Dell and HPE are credible alternatives without the governance baggage.

RVTY was considered as a life sciences diversifier but rated NEUTRAL at 55% confidence. Its connection to the AI creative destruction theme is too tangential to justify inclusion.

The Risks You Need to Take Seriously

Every ticker mentioned above carries real risk, and intellectual honesty about those risks is what separates analysis from cheerleading.

The single biggest risk across this entire thesis is that AI investment turns out to be a bubble. If companies realize they're spending more on AI compute than they're generating in AI revenue, capex slows down fast. Every infrastructure play listed here would get hit, with NVIDIA facing the steepest drop given its valuation.

China export controls affect NVIDIA, ASML, AMAT, and LRCX directly, shrinking their addressable markets and creating geopolitical uncertainty. Customer concentration is a recurring theme: Broadcom depends heavily on Google and Meta, Arista's top four customers are a massive share of revenue, and TSMC capex timing creates lumpiness across the entire equipment chain.

For the power plays, nuclear plants face operational and regulatory risks that are genuinely catastrophic when they go wrong. Interest rate sensitivity affects capital-heavy names like Constellation, Vistra, Equinix, and Eaton.

And nearly every stock on this list is already trading at valuations that reflect significant AI optimism. Stretched valuations mean even modest disappointments can trigger meaningful selloffs.

Why This Matters for Your Money

If you have a 401(k) or any investment account with exposure to a broad tech index fund, you're already making a bet on this pattern without realizing it. The creative destruction cycle means your tech fund's performance is being driven by a shrinking number of AI-focused mega-caps while the broader tech workforce contracts. The index looks fine, but the companies inside it are diverging sharply.

Understanding the shovel-seller framework gives you a way to think about AI investing without trying to predict which chatbot will be the best in 2026. The prediction markets are split on that question. What they're not split on is whether the race will require chips, networking equipment, server racks, cooling systems, and electricity. It will.

The creative destruction playing out in tech right now also touches everyday life. Those 87.1% layoffs translate into real people in your community. The AI systems replacing them will change how your doctor diagnoses illness, how your kids' homework gets graded, and how your company makes decisions. Whether that makes your grocery bill go up or down is still an open question. What's less open is that the infrastructure powering it all needs to be built, and the companies building it are the closest thing to a sure bet in an uncertain race.

Analysis based on prediction market data as of April 1, 2026. This is not investment advice.