10-Year Yield Near 5%, Yet Chip Stocks Keep Climbing

<div class="badge">MARKET SIGNAL</div>

Indexes rose, but the rally wasn't broad

<div class="metric-grid"> <div class="metric"><b class="up">+0.17%</b><span>S&P 500<br>7,650.50</span></div> <div class="metric"><b class="up">+0.39%</b><span>Nasdaq<br>26,522.55</span></div> <div class="metric"><b class="down">-0.18%</b><span>Dow<br>51,682.64</span></div> <div class="metric"><b class="down">-0.50%</b><span>Russell 2000<br>2,860.40</span></div> </div>

On the surface, Thursday, Sept. 18 looked like a quiet up day for US stocks. The S&P 500 and Nasdaq both edged higher, with the Nasdaq the stronger of the two. But the Dow and the Russell 2000 fell. The weekly tally tells a similar story: the Nasdaq gained 0.7% for the week, while the S&P 500 slipped 0.1%, the Dow dropped 1.7%, and the Russell 2000 lost 1.5%.

That combination doesn't look like a broad return of risk appetite. It looks more like selective risk-taking, with money concentrating in a narrow band of names where earnings expectations are strongest, chiefly AI infrastructure.

A 5% yield doesn't press on every growth stock the same way

The 10-year Treasury yield climbed to 4.998%, effectively touching 5%. That number carries psychological weight, but it also means the discount rate used to value future growth-stock earnings is rising again.

In that environment, companies whose profits are weighted toward the distant future, that depend heavily on outside financing, or that lean on revenue growth rather than cash flow tend to get hit first. The Russell 2000's 0.50% decline fits that pattern.

That distinction matters. Right now the market isn't treating growth stocks as a single basket. It's paying a premium for anything tied directly to revenue and order flow, AI data center buildout, memory pricing, leading-edge equipment, network bottlenecks, while judging growth stocks that trade purely on high multiples far more harshly.

Why chips propped up the market

SOXX closed at 533.07, up 2.69%. The more interesting detail is that the recent strength isn't confined to a single GPU name. It's spreading across the supply chain, into equipment, memory, and storage.

That points to a shift in how the market reads the AI investment cycle: less about the performance of a single chip and more about how much of the entire system still needs to be built. As more GPUs go into servers, bottlenecks cascade down the chain, into HBM, NAND, advanced packaging, optical networking, and power and cooling.

By contrast, mobile- and telecom-heavy chipmakers such as Qualcomm traded weaker the same day. Even within semiconductors, the premium now depends on how directly a company is tied to AI data centers.

Oil is still too high to relax about

Data sourcing matters here. Some real-time feeds showed Brent crude dipping below $100 a barrel, but Reuters' regular-session close for Sept. 18 put Brent at $104.87 and WTI at $100.30. It's premature to call oil "settled below $100."

Still, the direction, retreating from the week's highs, matters. If oil keeps falling, it could ease both inflation expectations and fears of further rate hikes. If it instead firms back up near $100, that would likely press on long-term yields and growth-stock multiples at the same time.

"Climbing a wall of worry" is only half right

The VIX fell 4.08% to 14.81. Options markets aren't pricing in a sharp near-term selloff. The CNN Fear & Greed Index, on the other hand, sat at 29, still in "Fear" territory.

That combination isn't panic. Cash equity investors remain cautious, but derivatives markets aren't flashing systemic alarm. In that kind of environment, the market is less likely to move as one block and more likely to see wide performance gaps between stocks based on earnings and valuation.

That argues against reading today's fear gauges as a simple contrarian buy signal. It's more useful to watch them alongside three things at once: rates, oil, and market breadth.

The stock-by-stock picture is clearer

</markdown>

Insight Times Editorial Desk