How to Harvest Semiconductor IV Without Getting Run Over

I own semiconductor stocks. Not a huge allocation. But enough that the last two weeks have been uncomfortable.

SMH closed Friday at $556.53. A month ago it was above $650. AVGO dropped from $394 to $371 in the last three trading days. AMD gave up $33 in the same stretch. NVDA is off 14% from its $236 high and hasn’t had a green week in a month. SMCI is SMCI. $24 after a $62 high, and that $7 billion capital raise in June didn’t help.

I didn’t sell any of it. I also didn’t buy the dip. What I did instead was check the options chain, and that’s what I want to walk through.

IV expanded. That’s the opportunity.

The VIX moved from 15 to almost 19 in two weeks. That’s not a panic spike, it’s a repricing. But for premium sellers, it’s the repricing that matters, because it shows up in the bid on puts and calls across the semiconductor complex.

Stocks drop, fear rises, option buyers pay up for protection, and sellers get a better price for taking the other side. You’re being compensated for catching a potential falling knife. The question is whether the compensation is worth it.

Separating noise from broken stories

NVDA at $202 is down 14% from its high. The AI capex cycle hasn’t reversed. Hyperscalers are still ordering GPUs. The stock is correcting, but the underlying demand for compute isn’t going anywhere. That’s a repricing, not a broken thesis.

AVGO is down 25% from $495. The company reported solid numbers and trades at a multiple that’s reasonable relative to its AI revenue growth. Sector rotation, not a company problem.

AMD at $495, down 15% on no obvious negative catalyst. Datacenter GPU ramp continues.

These aren’t broken companies. They’re expensive stocks in a sector that’s correcting. That distinction matters. You don’t sell puts on broken companies. You sell puts on good companies whose stock prices happen to be lower than they were.

SMCI is different. Down 61% from its high. Capital raise. Sentiment shift. I’m not selling puts on SMCI. The premium might be enormous, but the fundamentals are too uncertain for me to size a trade I’d be comfortable holding if assigned.

The mechanics

Take NVDA at $202.81. A 30-delta put 30 to 45 days out, the standard framework from tastylive’s research on optimal duration, lands around $185 to $190. That’s 6% to 9% below current levels in a name that’s already corrected 14%. The IV expansion means the credit on that put is wider than it was two weeks ago.

One NVDA put at $185 represents $18,500 in notional. In a $50,000 account, that’s 37% of your capital in one name. Don’t do that. In a $500,000 account, it’s 3.7%. The math doesn’t change because the premium looks good.

For smaller accounts, SMH at $556.53 is the cleaner trade. One instrument, a dozen names, diversified exposure. Sell a put 5% to 7% out of the money and you’re selling premium on the entire semiconductor sector without single-stock concentration.

What gets you

A correction can become a bear market. Semiconductors are cyclical. If the AI spending cycle slows, these stocks go lower. The VIX at 18 is elevated but not extreme. At 25 or 30, the premium you sold at 18 looks cheap in hindsight. That’s the trade.

I don’t know where the bottom is. Neither does anyone else. When IV expands and the underlying fundamentals are intact, I check one thing: has the price of insurance moved enough to make selling it worth the risk. On NVDA, AVGO, and AMD, the answer is yes.

Disclaimer: This is not financial advice. It’s not a recommendation to buy or sell any security. Options involve risk and aren’t suitable for all investors. I own semiconductor stocks. I may sell puts on them. You should make your own decisions based on your own research and risk tolerance.