Is the AI stock selloff finally over?

AI and semiconductor stocks have just come through a brutal correction, followed by a sharp two-day rebound. After weeks of selling in everything tied to AI infrastructure, the market finally bounced—raising the big question: was that the bottom, or just a pause before another leg down?

How leverage turned an AI correction into a crash

The recent AI-driven selloff wasn’t just about fundamentals. A huge part of the move came from leverage and margin getting washed out of the system.

In the U.S. alone, margin debt climbed to around $1.4 trillion. At the same time, leveraged products and aggressive hedge funds piled into complex positions around AI, semiconductors, and related tech names. When volatility picked up, that leverage became a problem.

One high-profile example was a young AI-focused hedge fund manager who reportedly turned a few hundred million into tens of billions at the peak—only to be forced into liquidation and ultimately bought out by Citadel. His fund was heavily short semiconductors and AI leaders (through positions in SMH, Nvidia, Oracle, Broadcom, AMD, and more) via puts opened back in March. Those shorts went against him as AI stocks ripped to new highs into April, May, June, and early July.

When the market finally cracked, he had to close those losing shorts while his long positions also dropped, triggering a vicious margin spiral. This story is likely just one of many. Leveraged ETFs tied to AI and tech reportedly saw assets fall from around $163 billion to $60 billion during the correction, and in Korea, roughly one in 30 investors faced margin calls.

The takeaway for individual investors: margin can turn a normal drawdown into a wipeout. For most retail traders, staying cash-secured and avoiding leverage can be the difference between riding out volatility and getting forced out at the worst possible time. If you’re wondering whether the AI boom itself is a bubble, it’s worth reading this deeper look at the $31 trillion AI stock wave.

What volatility is saying about AI and tech right now

To understand where AI and tech might go next, it helps to look at volatility indexes rather than just price charts.

The traditional VIX (based on the S&P 500) briefly pushed above 20 during the recent flush, but it didn’t spike to the 25–30+ levels you’d typically see in a full-blown panic. That made it harder to rely on VIX alone for timing entries in AI-heavy portfolios.

Instead, tracking VXN—the volatility index for the Nasdaq 100—gave a clearer picture. VXN climbed toward levels last seen during prior macro scares (like the Iran conflict and tariff tensions), reflecting how hard tech and AI names were hit. Many AI infrastructure stocks were down 20–50% from their highs, even as the broader market looked more stable.

Using volatility to guide cash allocation helped manage risk. With VIX in the 15–20 range for much of the move, keeping roughly 20–25% in cash allowed for buying dips in AI names at more attractive prices. Even when volatility spiked above 20, maintaining some cash (around 11% at the lows, then back up to 20% after the bounce) created flexibility for a possible second leg down.

That’s the current stance: cautiously optimistic that the bottom may be in, but still holding enough cash to take advantage if AI and tech retest recent lows.

Where the Nasdaq and AI stocks could go next

For AI-focused investors, the Nasdaq 100 (tracked by QQQ) is the key index to watch. Market makers are currently pricing in a wide expected move over the next couple of weeks, with potential upside toward the 715 area and downside toward roughly 668.

Two main paths stand out:

• Retest and grind higher: If negative news hits, QQQ could retest the recent lows around the 660s, forming a double bottom before grinding higher. That would likely mean another shakeout in AI and semiconductor names, but could offer even better long-term entry points.

• Steady recovery: If macro news stays neutral or positive, QQQ may continue to climb back toward its 20-day and 50-day moving averages, with AI and big tech leading the way.

Options flow has already started to turn more constructive. There’s been notable institutional call buying in Amazon, Microsoft, Nvidia, Google, semiconductor ETFs, and Alibaba, even as some hedging continues in names like Apple, Micron, and the S&P 500.

The current plan: stay positioned for more upside, but mentally and financially prepared for another dip. A retest of the 660 area on QQQ wouldn’t be a surprise and doesn’t necessarily mean the AI story is broken.

New AI-adjacent position #1: Apple for stability and buybacks

One of the new additions on this pullback is Apple. While not a pure-play AI stock, Apple is increasingly building AI into its ecosystem and offers something many high-flying AI names don’t: stability, massive cash flows, and relentless buybacks.

Apple’s latest earnings were strong:

• Revenue up about 16% year-over-year to $109 billion

• Earnings per share at $2.02, roughly 29% year-over-year growth

• iPhone, Mac, and Services all performed well

• Roughly $33 billion returned to shareholders, with Apple continuing to run one of the largest share buyback programs in the world

Despite that, the stock sold off after earnings and slid toward the lower Bollinger Band, with its P/E compressing from around 40 to roughly 34. That combination of strong fundamentals and technical weakness created an attractive entry point for a more defensive AI-adjacent holding.

The approach taken was to sell cash-secured puts rather than buying shares outright. Specifically, two August 295 puts were sold, bringing in around $1,120 in premium for roughly a 1.9–2% return on collateral over 28 days. With a delta near 0.35, this is an aggressive but intentional way to potentially get assigned shares a few percent lower while being paid to wait.

If assigned at 295, the plan is to build a longer-term position and potentially add more around 280–275 if the market gives the chance. Seasonally, Apple often does well into October and November, and its AI integration story (especially on-device intelligence and services) could become a bigger narrative into year-end.

New AI-adjacent position #2: Google at a rare discount

Google is the second new addition, and it’s much closer to the core AI race. Between Search, YouTube, cloud, and its deep AI research stack, Google is one of the key hyperscalers competing to monetize AI at scale.

What makes Google particularly interesting now is valuation. The stock is trading at around a 17x P/E—nearly half the multiple it commanded just months ago when it was closer to 31x. That means earnings have grown while the market has become more cautious, largely because Google is pouring huge amounts of free cash flow into AI infrastructure and capital expenditures.

Some investors are nervous about that spend, but the thesis here is that Google, alongside Microsoft and Amazon, is one of the few players with the scale and data to generate massive returns on AI investments. For a broader perspective on why large, cash-rich companies like Google are often seen as top AI picks, you can check out this breakdown of Google as a leading AI stock.

Instead of buying shares immediately, a starter position was opened via a single cash-secured put at the 340 strike, expiring August 28. That contract brought in about $710 in premium. The 340 level lines up with the volume-weighted average price (VWAP) area and is viewed as a fair value zone. If assigned, the plan is to hold and look for a move back toward the 400 region over time.

Managing GLW after a deep AI infrastructure selloff

Corning (GLW) is another key AI infrastructure play in this portfolio. While it’s not an AI software name, it’s critical to the physical backbone of AI: glass, fiber optics, and materials used in data centers and high-speed networking.

GLW’s valuation has reset dramatically. The stock now trades at a P/E around 64—the same multiple it had when the share price was near $100 back in January. The difference is that the stock price is now closer to $140 while earnings per share are much higher. That disconnect stems from a sharp selloff, even as fundamentals have improved.

Recently, Corning announced a multi-billion-dollar partnership with Amazon to expand production capacity for glassware and fiber optic cables—exactly the kind of infrastructure AI workloads depend on. Despite that, GLW fell as much as 50% from its highs during the AI correction.

In this portfolio, shares were assigned at an average cost basis around $189, with roughly 800 shares held. To manage the position:

• Some lower-strike puts (around 135) that were unlikely to be assigned were closed into strength.

• If GLW dips back toward 135–136, the plan is to buy 200–300 additional shares outright to bring the cost basis down into the mid-170s.

• From there, covered calls dated out to late August at strikes around 170–175 can generate income while waiting for a recovery toward the VWAP area (roughly 160–170).

Technically, GLW remains oversold on RSI, with a recent bullish crossover and a MACD that’s close to turning higher if strength continues. Earnings looked solid, and the long-term AI infrastructure story remains intact, so the strategy is to average down carefully and monetize volatility with covered calls.

Managing DRAM: concentrated exposure to AI memory

DRAM, a memory-focused ETF, is another large position that was hit hard during the selloff. It holds key players in the memory and storage ecosystem, including SK Hynix, Micron (MU), Seagate (STX), Western Digital (WDC), and SanDisk. These companies are central to AI because modern AI models are incredibly memory- and bandwidth-intensive.

After a steep drop, DRAM is viewed as significantly oversold. The portfolio was assigned a substantial number of shares at various levels, including around $61 and $63, with an additional 400 shares set to be assigned at $56.50. That should bring the blended cost basis down into the high-50s.

To manage risk and generate income, covered calls have been sold on all shares, with strikes around 61.5 and 65—above the main assignment prices. This approach aims to capture both:

• Premium income: Regular option income while the ETF stabilizes and recovers.

• Upside participation: Some capital appreciation if DRAM continues to rebound from the lows.

The conviction here is that memory is a core bottleneck and enabler for AI workloads. As AI adoption grows, demand for high-bandwidth memory and storage should remain strong, making DRAM an attractive way to gain diversified exposure to that theme after a big reset.

Putting it all together: a balanced AI strategy after the selloff

The recent AI and semiconductor selloff was amplified by leverage, margin calls, and forced liquidations, but it also created opportunities. The current positioning reflects a balance between offense and defense:

• Defense: Avoiding margin, keeping 20%+ in cash, and adding more stable names like Apple at better valuations.

• Offense: Leaning into discounted AI and AI-adjacent names like Google, GLW, and DRAM via cash-secured puts and covered calls.

• Risk management: Using volatility (VIX and especially VXN) to guide cash levels and avoid overcommitting at any single point in time.

From here, the base case is that the worst of the AI stock selloff may be behind us, with a good chance of grinding higher into year-end. But another dip—especially a retest of recent lows on QQQ—wouldn’t be surprising and may offer even better long-term entries.

For individual investors, the key lessons are simple: respect leverage, stay flexible with cash, and focus on high-quality AI and infrastructure names that can compound over years, not just weeks.