Chip Stocks Bear Market: Should You Sell Now?

Chip Stocks Bear Market: What Changed and What Didn’t

Weekly Market Review for the week ended July 18, 2026. Adapted from the Berk on Value video.

Should you sell chip stocks after this week?

The question a real client asked on Tuesday, after IBM’s worst day since 1987.

Our answer was no. Three reasons, none of them a prediction.

First, the evidence. Outside of IBM, not one company in the selloff reported bad news. AMD, Taiwan Semiconductor, Broadcom, Micron, Intel: all fell hard, none warned. IBM’s warning is evidence about IBM’s mainframe cycle, not about foundries running at record utilization.

Second, the sizing. Positions were built so a 20% single-sector drawdown never forces a sale. That is the whole point of sizing: the decision gets made before the drawdown, not during it. We also hold cash so a client never has to sell a great business at the worst moment.

Third, the falsifiers. We are not stubborn; we are specific. Three pieces of evidence would change the answer: TSMC or the big cloud builders cutting capital-spending guidance instead of raising it, Kimi K3’s efficiency benchmarks replicating under independent testing at scale, or order cuts showing up in actual chip company reports. None of the three happened this week. TSMC did the opposite.

Everything below is the evidence behind those three sentences.

The scoreboard: a narrow selloff

Did the market fall this week, or did a few stocks fall?

A few stocks. The S&P 500 (SPY) closed Friday at 743.29, down 1.5% for the week and still up 8.8% for the year. The equal-weight version (RSP) slipped 0.4%. When the cap-weighted index falls almost four times as far as the equal-weight index, the selling sits in a handful of giant names. The Nasdaq 100 took the real hit, down 4.2%, as the tech selloff deepened while energy stocks rose on the oil spike. Small caps barely moved, down 0.7%, and the VIX closed at 18.8, up about 25% for the week.

The sector spread was wide. Energy gained 4.7% as crude surged and now leads the whole market for 2026, up 29%. Real estate and consumer staples came next: the classic defensive playbook. Technology sat at the bottom, down 5.5% in five sessions, with consumer discretionary and industrials also lower, a pattern Morningstar’s weekly recap put plainly: energy rose while technology fell.

The finding that matters: money did not leave the market this week. It moved out of crowded AI names and into energy and defensives.

The decision this forces is smaller than it feels. A 4-point gap between the Nasdaq-100 and the equal-weight S&P is a reason to check whether your tech weight drifted above plan, and rebalance if it did. It is not a reason to exit the market that just showed you its average stock is fine.

Two events cracked the AI trade

Why did chip stocks sell off when no chipmaker reported bad news?

Two headlines changed assumptions, not order books. The PHLX Semiconductor Index closed more than 20% below its June peak, the formal bear-market line, and about $3.3 trillion of global chip market value has been erased since June 22. Hold that scale against this fact: the selling traces to exactly two pieces of news, neither from a chipmaker.

First, IBM. On Monday, the company pre-announced its quarter: revenue up just 1%, the mainframe cycle stalled, big deals slipping. CEO Arvind Krishna admitted, “We did not adapt and move quickly enough”. The stock lost about a quarter of its value in a single day. The full report lands July 22, so investors get the detail behind the warning within days.

Second, Kimi K3. Moonshot’s new model claims frontier results with less computing hardware than US spending plans assume, and markets treated it as a new DeepSeek shock. Jim Cramer put it bluntly on Friday: “There will always be another DeepSeek.”

A client called on Tuesday after the IBM news and asked whether to trim every tech name in the portfolio. The answer was no. IBM’s warning is evidence about IBM. It says little about companies whose demand just hit records, as the TSMC section below shows.

The economy quietly improved underneath the drama

What did the week’s economic data actually say?

It said the backdrop got better. June CPI fell 0.42%, the first monthly decline since 2020, with the core gauge unchanged. Annual inflation came in at 3.5%, less than expected as energy prices eased. Consumers kept spending: June retail sales rose as lower gas prices freed up money, with sales excluding gas up 0.7%. Jobless claims hit a ten-week low of 208,000, and the University of Michigan consumer sentiment index reached a five-month high of 54.4. The retail control group, the slice that feeds GDP, ran at a 9.2% annualized pace for the second quarter.

Then the Fed. New chair Kevin Warsh told Congress the central bank has “no tolerance” for persistently elevated inflation. Markets read patience in his hands anyway: the CPI miss took a July hike off the table, and futures now price just one hike over the next twelve months.

Cooling inflation, a spending consumer, and a patient Fed make for a friendly backdrop for business owners. One caution flag: core producer prices are still running near 4.8%. The inflation fight is not finished.

Oil, Iran, and China set the tone

Why does a shipping lane in the Persian Gulf show up in your gas price?

Because about a fifth of the world’s oil moves through the Strait of Hormuz. Iran’s Revolutionary Guard fired on ships there, Tehran declared the strait closed, and US forces struck targets inside Iran on Friday. Brent finished at $88.10 a barrel, up 15.9% in a week and 44.8% this year. Every dollar of that flows into oil producers’ earnings. It also feeds back into gas prices and inflation, which is exactly why the Fed stays careful.

China added two more jolts. Its economy grew by 4.3%, the slowest pace since late 2022, while record exports continue to push prices down worldwide. And its newest AI model shook Wall Street. Geopolitics set oil prices this week, and oil set the tone for half the market.

Winners and losers: one story, ten tickers

What do the week’s biggest movers have in common?

They are the same story wearing different tickers: money repricing AI risk and rewarding whatever benefits from it.

The market’s movers first. The two best S&P 500 performers were cybersecurity companies: Palo Alto Networks jumped 10.1%, and CrowdStrike rose 8.5%, because cheap, capable AI models multiply the attackers corporate networks must stop. Chevron gained 6.2% and Exxon 6.1% on the oil spike. Apple rose 5.8%, set a record high, and briefly passed Nvidia as the world’s most valuable company. A chip bear market and an Apple record in the same tape is the rotation signature: an exodus takes everything down together.

The biggest S&P losers after IBM’s 26% collapse: Intel down 13.5%, Micron down 13.3%, AMD down 11.1%, Broadcom down 7.3%. None of the four reported any bad company news.

The portfolio’s best. Diamondback Energy led, up 6.6% for the week and 30.1% this year, a Permian producer whose barrels got more profitable by the day. Accenture rose 6.2% after falling by nearly half this year amid fears that AI will replace consultants, even as revenue and earnings keep growing; it is still down 46.5% in 2026. Apple added 5.8% to reach that record, up 22.8% this year. S&P Global gained 4.7%, a toll road collecting fees on ratings, indexes, and data, still down 13.7% for the year. KKR rounded it out, up 4.1% on a joint venture for Thomson Reuters’ print business, still down 20.8% in 2026. Three of these five remain down big for the year. We show that on purpose: winners and losers in the open, every week.

The portfolio’s worst. AMD dropped 11.1%, Taiwan Semiconductor 8.2%, Broadcom 7.3%, Tesla 6.6%, ServiceNow 4.2%. Every one is the same story wearing a different ticker. AMD is still up 131.5% this year even after the drop. Taiwan Semi is up 31.1% in 2026. Tesla just posted record deliveries of 480,126 vehicles and reports on Wednesday. ServiceNow keeps growing revenue near 20% while the market prices it for AI extinction, down 32.6% this year.

What changed was the mood. Mood is not evidence. Positions were already sized for this kind of volatility, so we hold, we watch Wednesday’s reports, and we let the businesses speak.

The most important report of the week: TSMC

What happens when record results meet a falling price?

Taiwan Semiconductor manufactures the advanced chips for Apple, Nvidia, and AMD. On Thursday, it reported a record quarter: revenue of $40.2 billion, net profit up 77%, guidance raised to above 40% growth, and another $100 billion committed to Arizona, taking planned US investment near $265 billion. CEO C.C. Wei said AI demand stays strong through the end of the decade.

The stock fell 8.2% anyway.

Record results, raised guidance, falling price. When price and value diverge like that, long-term owners pay attention. For what it’s worth, 20% chip drawdowns earlier in this cycle resolved higher, though history is context, not a promise. Weeks like this are where future returns get planted.

Put plainly: the market repriced TSMC’s customer list. It did not reprice TSMC’s order book. Krishna’s warning and Moonshot’s benchmarks are claims about who buys chips and how many. The $40.2 billion quarter and raised guidance are the current answer, and this week only one side of that argument showed up with audited numbers.

Next week: eight reports and two paths

What should you actually watch?

The busiest stretch of the quarter. General Motors leads off Tuesday, July 21. Wednesday, July 22 is the marquee day: Alphabet and Tesla headline, with ServiceNow and Medpace alongside. Thursday, July 23 brings Roper, Kinsale, and Thermo Fisher. Beyond earnings: existing and new home sales, the manufacturing and services PMI surveys, and the Fed’s July 28-29 meeting with a policy decision and press conference.

• The constructive path: Alphabet and Tesla deliver solid numbers with confident AI commentary, chips stabilize after a 20% drawdown, oil settles, and the rally broadens out.
• The darker path: a second wave of guidance disappointments confirms IBM was the first domino, Kimi K3’s benchmarks hold up and force hardware forecasts lower, and Brent pushes toward $100 right before the Fed meets.

Which path wins probably comes down to two earnings reports Wednesday night. We don’t predict. We prepare.

How long do chip bear markets last?

The follow-up question every selloff raises, answered without a forecast.

It depends on which kind this is, and nobody knows yet. Within the current AI cycle, 20% chip drawdowns have resolved higher, usually within months. Full semiconductor downturns are a different animal: past cycle busts cut the group 30% to 80% and took years to round-trip. Anyone who tells you which one this is has skipped the part where the evidence arrives, starting with Wednesday’s earnings.

Our answer is to stop guessing the duration and remove its power. Size every position so the deep case does not force a sale. Hold cash so nothing has to be sold at the bottom. Then the only question left each week is the one this review runs on: did the evidence about the business change? That filter does not shorten a bear market. It makes the length survivable, which is the only part an investor controls.

Key takeaways

• The AI trade got stress tested: price corrected while demand did not. TSMC’s record quarter landed the same week the chip index entered a bear market.
• The macro backdrop quietly improved, with the first monthly CPI decline in six years and a patient Fed. Oil’s 16% surge says the inflation fight isn’t over.
• Rotation is healthy. Equal weight beat the cap-weighted index, and small caps lead the year. The market is gaining breadth, and breadth is strength.

Before you react to anything next week, run it through one filter: did the evidence about the business change, or just the mood?

This post was adapted from the Berk on Value weekly market review video. Watch the full breakdown here: