S&P 500 Near Record Highs — But Half the Market Is Quietly Drowning

Okay, so this one actually surprised me.

I was looking at the S&P 500 this morning — sitting within 2 percentage points of its all-time closing high, looking smug as ever — and then I caught the MarketWatch piece that pulled back the curtain. Nearly half of the individual stocks inside that index are in a bear market right now. We're talking 20% or more below their recent peaks. At the same time the headline number is practically printing new records.

That is a genuinely strange thing to have happen. And I think it matters more for your actual portfolio than most people realize.

What the Index Number Is Hiding

The S&P 500 is a market-cap-weighted index. That sounds boring, but here's what it means in plain terms: a company worth $3 trillion counts thousands of times more than a company worth $2 billion. When Nvidia, Apple, Microsoft, and a handful of other mega-caps are on a tear — which they largely have been, riding the AI capital expenditure wave — the index can float higher even if hundreds of the smaller names inside it are getting quietly crushed.

CNBC reported today that almost half of S&P 500 components now carry a negative beta. That's the stat that stopped me cold. Beta measures how a stock moves relative to the broader market. A negative beta means the stock is actually moving against the market's direction. Not just underperforming — moving in the opposite direction. When nearly half the index is doing that, you don't really have a "market" in any meaningful sense. You have a small cluster of heavyweights pulling the rope one way while hundreds of other stocks get dragged backward.

Why This Is Happening Right Now

Rising Treasury yields. Full stop.

That's the thread running through almost everything strange about this market in late September 2026. Yields have been climbing steadily — the Dow dropped more than 300 points Monday as yields continued their ascent — and that's not a neutral background event. Higher rates are a gravitational force. They pull money out of equities that don't have ironclad earnings power, and they punish anything that's priced on future promises rather than current cash flows.

The companies that are surviving — thriving, even — are the ones that can credibly say: "We're making real money right now, and AI demand is making sure we make more of it tomorrow." That's your Nvidias, your memory suppliers like Micron, your hyperscaler-adjacent names. They've got the pricing power and the CapEx tailwinds to fend off rate pressure. Micron, for instance, is generating profit margins this cycle that would have seemed almost fictional in any previous memory cycle. Nvidia's trading at roughly 24x earnings — below its five-year average — while the AI infrastructure buildout keeps printing demand.

Here's what I actually think is happening: the bond market is functioning as a sorting mechanism. It's separating companies with genuine earnings from companies that were just along for the ride. And the ride, apparently, is over for a lot of them.

This Isn't Just an Academic Observation

Let me put this differently. If you own a broad S&P 500 index fund, you might look at your account, see it's basically flat to slightly positive, and think: "Cool, market's fine." What you might not see is that inside that fund, your slice of Nvidia or Microsoft is doing the heavy lifting while fifty other positions you own are quietly leaking value.

That's not catastrophic — index funds are still the right vehicle for most long-term investors, and dollar-cost averaging into them still works even in choppy conditions. But it does mean the "diversification" you think you're getting from 500 names is a bit of an illusion right now. You're more concentrated in a handful of AI-adjacent mega-caps than the name "S&P 500" implies.

And this is where I think most people get it wrong — they confuse index exposure with broad market exposure. Those two things are supposed to be roughly equivalent. Right now, they're not.

The Bonds Angle Is Worth Taking Seriously

Robin Wigglesworth over at the FT has been making the rounds talking about what bonds are actually signaling under the hood — about inflation, about government debt, about where the real stress in the system is building. And honestly, the bond market has been the canary here for months. Treasury yields don't surge in a vacuum. They're expressing something: either growth expectations, inflation fears, or some combination of demand destruction at the long end of the curve where nobody wants to park money for 10 or 30 years without serious compensation.

The practical translation is this: when the 10-year yield climbs, it doesn't just hurt bond prices (which is why rate-sensitive ETFs are getting hammered right now — CNBC flagged two in particular "breaking down" from key macro trades). It also recalibrates the discount rate that investors apply to every future dollar a company might earn. Small growth stocks, regional banks, anything with a long earnings runway — they all get hit with a higher hurdle. The math just works against them in a high-rate world.

The mega-caps escape this partly because their earnings are so massive and so near-term that the discounting effect is muted. They're not speculative futures. They're NOW money.

How Bad Is "Bear Market" for Half the Index?

Worth being specific here, because "bear market" can sound alarming in ways that need grounding. A stock is technically in a bear market when it's down 20% or more from a recent high. That's not the same as a company going bankrupt or the economy collapsing. It might mean a perfectly fine company with decent fundamentals got caught in a rising-rate rotation and repriced downward.

But it also might mean the earnings story never materialized, or AI automation is quietly undercutting a business model, or the company got too expensive during the 2024–2025 everything rally and is now just reverting to reality. Probably some of each, honestly.

The labor market data isn't helping the less-glamorous corners of the market either. Job openings are low, hiring is weak, and a piece out today from MarketWatch makes the case that war, energy prices, higher rates, and — yes — AI are capping job creation in ways that won't reverse quickly. A soft labor market usually feeds into softer consumer spending, which is particularly rough for the mid-tier consumer-facing companies that lack AI narrative cover. They're bearing the brunt of rate pressure without the offsetting tailwind.

Look, I could be wrong here, but I think we're in a market where the "headline number is fine" story is going to hold for a while — as long as the AI CapEx cycle keeps feeding the mega-caps. The problem is that a market priced that narrowly is fragile. If one or two of the names carrying the whole index stumble — and Micron's upcoming earnings this week will give us a real-time read on whether the memory boom has more runway — the cushion disappears fast.

What's Propping the Headline Number Up

A quick look at where the weight actually sits helps explain the persistence of the S&P 500 near its highs even while breadth is this bad.

The top 10 holdings in the index have routinely represented 30–35% of total index weight in recent years. In a world where those 10 names are mostly riding the AI infrastructure wave — and where hyperscalers are still showing stubborn CapEx appetite — they can absolutely drag the whole index reading higher even if names 200 through 500 are in various states of decline.

There's a useful historical parallel here: the dot-com era showed us how long a narrow, cap-weighted index can defy a crumbling breadth picture. From mid-1999 through early 2000, the Nasdaq was printing new highs on the back of a handful of names while the majority of technology stocks were already rolling over. It didn't end well when the load-bearing names finally cracked. I'm not saying that's where we are — the earnings powering today's leaders are real in a way dot-com revenues largely weren't — but the structural dynamic of index concentration masking underlying weakness isn't a new story.

This is part of what the CAPE ratio keeps flagging as a concern at the aggregate level — when index valuations are elevated and the concentration is this high, the margin of safety for the overall market is thin even if the top names individually look defensible.

Let's Talk About What This Means Practically

If you're a long-term, set-it-and-forget-it index investor: you don't need to panic. This breadth divergence can persist for longer than feels comfortable, and trying to time it is usually a losing game. But it's worth knowing that your S&P 500 index fund is functionally a bet that AI mega-cap dominance continues — not a diversified bet on the broad US economy.

If you're actively picking stocks or managing tilts in your portfolio: this is genuinely valuable signal. The bond market is telling you that rate-sensitive sectors remain under pressure and that the "rising tide" narrative is off the table. You need real earnings power, pricing strength, or some direct connection to the AI buildout to outperform right now.

And if you're looking at "cheap" stocks that have been beaten down 20–30% this year — which the breadth data suggests there are plenty of — the question to ask isn't just "is it cheap?" It's "why hasn't the market already bid this back up, and has anything about that reason changed?" Sometimes the answer is yes. Sometimes it's a value trap wearing a discount sticker.

What to Watch Before You Make a Move

• Micron earnings this week — if margins hold at current elevated levels, the memory/AI infrastructure trade has legs and may pull other tech names higher. If they miss, watch whether the mega-cap AI names decouple or get dragged down too. That answer tells you a lot about how structural the breadth problem really is.
• The 10-year Treasury yield — if you're holding rate-sensitive ETFs or financials, track whether yield momentum stalls near current levels. Traders are already watching for this; CNBC noted today that at least one of the beaten-down macro ETFs is being positioned for a bounce if yields stabilize.
• S&P 500 equal-weight index (RSP) vs. the cap-weighted SPY — this spread is the cleanest real-time read on whether breadth is improving or deteriorating. If RSP starts closing the gap with SPY, the rotation is happening. If it keeps falling behind, the concentration trade is deepening.
• For long-term index holders: nothing urgent to do, but check your actual allocation. If you hold both a total market fund and a large-cap growth fund, you may have far more overlap — and far more mega-cap concentration — than you think.
• For anyone considering adding to beaten-down names in retail, housing, or industrials: the thesis needs a rate catalyst to really work. Watch the Fed's next commentary signal before sizing up that position.