94.5%: The Fed Hikes Tonight — And Your Semiconductor ETF Is Already in a Bear Market

By Yogurt · 2026-09-16 · Market Analysis

The CME FedWatch Tool shows a 94.5% probability that the Fed raises rates 25 basis points tonight. But the hike itself isn't the story — the story is that the semiconductor sector is already down 24% from its $655 peak, a head-and-shoulders pattern is forming on the SMH, and Kevin Warsh's press conference could unlock another 50–75 basis points before year-end. Here's what every investor needs to know before the decision drops.

There are days when the headline number is the story. And there are days when the headline number is the distraction. Wednesday, September 16, 2026 is the latter. The CME FedWatch Tool shows a 94.5% probability of a 25-basis-point rate hike when the Federal Open Market Committee releases its decision this evening — the highest certainty reading of the entire current cycle. Markets have fully priced this. The hike itself is not news. What matters is the 47 minutes of questions that follow, when Federal Reserve Chairman Kevin Warsh stands at that podium and tells the world whether tonight's move is the last one or the first of several.

The Number That Actually Matters: 94.5%

The CME FedWatch Tool has been wrong before, but not often when the probability crosses 90%. At 94.5% — reading from the fed funds futures market as of Tuesday's close — the market is not betting on a hike. It is assuming one. The last time conviction was this high heading into a meeting, the Fed delivered. And the last time they surprised in the dovish direction at this probability level, the selloff in bonds was severe enough to make the pause feel like a mistake within weeks.

So the hike is happening. The 2-year Treasury yield, sitting at 4.67% and its highest point since 2023, has already processed this. The 10-year is also at a level not seen since late 2023. These numbers are not reacting to tonight's news — they are leading it, and they are telling a specific story: the bond market believes the Fed is not done after tonight.

How much not done? Bank of America's rate strategy team has put a number on it: 75 total basis points of hikes by year-end. Tonight accounts for 25. That implies two more after this one — in a calendar that has two more meetings before December 31. Barclays is slightly more conservative, penciling in 50 additional basis points beyond tonight. Goldman Sachs has also revised its terminal rate estimate upward in recent weeks. The herd has moved, and the herd is pointing in one direction: rates are going higher than anyone thought two months ago.

This is the scenario where the press conference is worth more than the decision itself. What Chairman Warsh says about the pace of future hikes — whether he leaves the door open, emphasizes data dependence, or leans into the inflationary persistence that the bond market is pricing — will set the tone for the next two meetings. A hawkish press conference into an already-priced hike can move markets further than the hike itself. Watch the 10-year yield at 2:30pm ET as your real-time readout.

Semiconductors: -24% From the Peak and Forming a Dangerous Pattern

While the Fed debate dominates the conversation, the real damage is happening in semiconductors. The SMH — the VanEck Semiconductor ETF, the broadest proxy for the sector — peaked at approximately $655 during the earlier phase of the AI infrastructure build-out. As of Tuesday's close, it has shed 24% from that high. That is not a correction. By every technical and historical definition, the semiconductor sector is in a bear market.

The chart is concerning for a specific reason: pattern recognition. With the 150-day moving average as the reference line, the SMH's price action over the past several months is tracing what technical analysts identify as a potential head-and-shoulders formation. For anyone unfamiliar with the pattern: a head-and-shoulders top is a reversal structure that consists of three peaks — a left shoulder, a higher head, and a right shoulder at roughly the left shoulder's level — followed by a breakdown through the neckline. It is one of the more statistically reliable bearish reversal patterns in technical analysis, and it is visible in the current semiconductor ETF chart.

The two scenarios playing out in real time: the bearish case points to a measured move target near $461 — the mathematical projection from the head-and-shoulders neckline break. The bullish case requires the SMH to hold above $537–541 and reverse higher from there, which would invalidate the pattern and suggest the lows are in. Tuesday's session brought the price close enough to that support range that the next 48 hours become determinative. If the $537–541 zone holds through Wednesday's close, the bull case survives. If it fails, the next significant support is materially lower.

Here is what makes the semiconductor bear market particularly complicated: it is not primarily a rate story. The sector's fundamental thesis — AI chip demand driving a multi-year spending cycle — has not changed. NVIDIA's business is not broken. Taiwan Semiconductor's order book is not empty. But the market is pricing something different: the pace of the AI infrastructure build-out has slowed, or at minimum paused, as hyperscalers reassess their capital expenditure timelines. The headline number from the AI capex wave — $500 billion in announced spending across Amazon, Google, Microsoft, and Meta — is real. But those dollars are flowing over multiple years, and the market is now asking what happens in the quarters between major deployment cycles.

Jensen Huang spoke at a Salesforce technology conference on Tuesday. His message was characteristically bold: every company on earth will become an AI company. He pushed back on any suggestion that AI development needs to slow down. But here is the market's response to Jensen Huang's confidence: SMH down, semiconductors in a bear market, SOXX declining toward levels last seen before the AI rally began. The market is not always right. But it is sending a signal that a single CEO's conference appearance is not enough to reverse.

Meta's Breakout Is the Day's Bright Spot

Not everything is red. Meta Platforms (META) staged a notable technical move on Tuesday, breaking above the declining trendline that has been capping the stock for weeks. The trigger: the release of Muse, Meta's latest AI model, which represents the company's most significant generative AI push since LLaMA. As of Tuesday's close, META was trading at approximately $670, with the next meaningful resistance at the $686 level.

For investors who have been watching Meta's positioning in the AI race, the Muse release matters for a structural reason. Meta has been the AI race participant that was consistently behind — OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude all preceded Meta's conversational AI offerings. Being late in AI has a cost: distribution, partnerships, and developer mindshare tend to consolidate quickly. The question is whether Muse represents a genuine capability jump or another release that closes the gap without clearing it.

The market's initial verdict — a breakout above the declining trendline — is modestly encouraging. The $686 resistance is the level that matters for confirmation. If Meta clears $686 on meaningful volume in the next few sessions, the stock is likely making a run at new highs. If it stalls and reverses there, the breakout was a dead-cat bounce and the trendline test resets lower. Watch the volume on any move toward $686 as the distinguishing variable.

The Clarity Act Fails — Bitcoin Falls Below $76,000

Tuesday's other major market event arrived from Capitol Hill: the Clarity Act, the proposed legislative framework that would have established clear regulatory rules for the cryptocurrency industry — including which tokens qualify as securities, what banks and financial institutions can hold, and how exchanges must register — failed its House vote. The defeat was bipartisan in the worst sense: every Democrat voted against it, and three Republicans crossed the aisle to join them, providing the margin that killed the bill.

The crypto market's reaction was immediate. Bitcoin, which had been holding above the $76,000 support level with notable consistency over the past several weeks, broke through that level on the Clarity Act news and was trading near $75,740 as of Tuesday evening. That $76,000 zone had been a reliable floor; the break below it removes a structural support that bulls had been pointing to as evidence of a base. The next significant support sits near $75,600. A close below that level would suggest the correction is accelerating.

The Clarity Act's failure matters beyond crypto. It is a signal about the regulatory environment's appetite for formalizing new asset classes — and that signal has implications for the broader fintech sector, for bank crypto custody plans, and for the institutional capital that had been considering increased digital asset exposure under a regulated framework. Without regulatory clarity, the institutional on-ramp remains narrow, and the speculative premium on crypto assets stays volatile. Watch Bitcoin's next 48 hours closely: the $75,600 level is the one that matters.

The Sector Rotation Nobody Is Talking About

Tuesday's session illustrated the rate-hike rotation with unusual clarity. Energy rose 2% on a combination of Middle East tension — Iranian threats around the Strait of Hormuz, Saudi production discipline, and geopolitical risk premium — and the simple reality that energy companies generate more free cash flow when rates are high than when they are low. High-yield, real-asset businesses with pricing power look relatively attractive when the cost of capital is rising, because they are less dependent on terminal-value assumptions that compress under high discount rates.

Consumer discretionary fell 0.75%. The mechanism is straightforward: rate hikes make mortgages more expensive, auto loans more expensive, and credit card balances more expensive. Households under that kind of pressure cut discretionary spending first — the Amazon Prime orders, the restaurant meals, the clothing budgets. The stocks that carry that consumer spending — Amazon's retail segment, apparel companies, durable goods manufacturers — trade the rate-hike narrative inversely. Tuesday's 0.75% decline in the consumer discretionary sector is not a one-day event; it is a repricing of expected earnings under a higher-for-longer rate assumption.

The Russell 2000 fell 0.96%. Small-cap companies are the most rate-sensitive segment of the equity market, because they carry proportionally more floating-rate debt, have less pricing power, and rely more heavily on credit markets that tighten as rates rise. The Russell's underperformance relative to the large-cap S&P 500 on Tuesday is the textbook rate-hike trade. If the Banks of America's 75-basis-point forecast proves correct, the Russell's underperformance through year-end will likely be substantial.

The S&P 500 Is Not Breaking — Yet

Amid all of this, the S&P 500 is behaving with relative composure. The index is trading within its established channel, approximately 3% below its recent all-time high, and remains above its 150-day moving average — the long-term trend line that separates bull-market pullbacks from something more structural. A 3% drawdown from an all-time high happens, on historical data, approximately seven times per year in an ordinary bull market year. In that sense, the current S&P level is not alarming.

What would be alarming: a break below the 150-day moving average on above-average volume after a hawkish press conference. That sequence — hawkish Warsh, bond yields ripping higher, defensive rotation, large-cap momentum names selling — is the scenario the bears are waiting for. It has not happened yet. Whether it happens tonight depends almost entirely on what Warsh says and how the 10-year Treasury yield responds in the 60 minutes after the press conference begins.

The Fear and Greed Index, CNN's sentiment composite that aggregates seven market signals into a single reading, closed Tuesday at 29 — in "Fear" territory and approaching the "Extreme Fear" threshold of 25 below which historically has marked buying opportunities. The index is not a trading signal; it is a sentiment map. At 29, the market is not euphoric. But it is also not at the capitulation readings that preceded the S&P's strongest recoveries. The inflection point — if there is one — comes after the press conference, not before it.

What to Watch Tonight and Tomorrow

2:00pm ET — Fed rate decision. The 25-basis-point hike is priced at 94.5%. The announcement itself is a non-event unless the Fed surprises with a hold or a 50-basis-point move, either of which would be genuinely shocking. Assume 25bp and focus entirely on what follows.

2:30pm ET — Kevin Warsh press conference. The language is everything. Watch for: (1) whether Warsh uses the phrase "meeting-by-meeting" — a dovish tell that signals conditional pausing; (2) whether he references the labor market as still "resilient" — a hawkish tell that justifies further hikes; (3) how he characterizes the dot plot's terminal rate projection. A hawkish press conference with the hike fully priced could send the 10-year yield toward 5% within the session. A dovish pivot — "we've done a lot, we'll wait to see the impact" — would likely ignite a bond rally and equity relief trade.

SMH $537–541 support zone. This is the technical inflection for the semiconductor sector. Hold here through Wednesday and the head-and-shoulders bear case is suspended. Break here and the $461 measured move target becomes the next conversation.

Bitcoin $75,600. A close below this level would remove the remaining near-term support and open the door to a retest of the $72,000–74,000 range. A close above $76,000 would suggest Tuesday's Clarity Act-driven flush was a one-day event rather than the beginning of a broader breakdown.

META $686. The level that confirms whether Tuesday's breakout above the declining trendline has institutional follow-through or fades back into the range.

The Only Thing That Changes the Semiconductor Thesis

A rate-driven selloff in semiconductors is uncomfortable but recoverable. The companies are structurally intact. The AI buildout is real, the capex commitments are documented, and the demand for advanced chips is not hypothetical. What would be genuinely damaging to the semiconductor thesis is a signal from the major hyperscalers — Amazon, Google, Microsoft, Meta — that they are decelerating their AI infrastructure spending. That signal has not come. Until it does, the 24% drawdown in the SMH is a cyclical correction in a secular bull story.

The question for investors sitting in semiconductor positions right now is the same one the market is asking: are you there because you want near-term gains, or because you believe the AI infrastructure thesis is multi-year? If the former, the next 48 hours are uncomfortable and the chart is not friendly. If the latter, the $537–541 support zone is the level to watch, the 150-day moving average is the line that matters for the broader bull case, and tonight's Fed press conference is the catalyst that either accelerates the selloff or provides the relief that lets the sector stabilize.

Either way, the answer is coming tonight. At 2:30pm ET, Kevin Warsh will step to the podium and say things that move markets. When he is done speaking, the next chapter of the semiconductor trade will have a new first sentence.