AI trade faces earnings test after Friday's chip rout; Tesla, Meta and GTC set the tone this week

The AI trade heads into an earnings gauntlet: after Friday's rout in chip stocks, Tesla, Meta and Nvidia's GTC take the baton this week^[1][2]. US tech sold off across the board on Friday, October 16, led by semiconductors and high-flying software names, with Apple the only gainer on the watchlist; more than 70 S&P 500 companies report this week, and investors will be testing whether AI infrastructure spending is converting into profits^[1][2].

Friday's tape: chips and software lead the decline, Apple alone rises

At Friday's close, AMD and Intel plunged 5.95% (to $528.75) and 5.79% (to $89.85) respectively, while Nvidia slipped 1.43% to $214.13; emerging AI software names bore the brunt, with Snowflake down 8.43% at $306.84, Palantir down 7.22% at $186.03 and C3.ai down 6.91% at $11.32^[1]. Among megacaps, Meta fell 3.72% to $709.33, Alphabet 2.47% to $340.59, Tesla 2.13% to $354.72, Microsoft 1.93% to $529.05 and Amazon 1.87% to $250.00 — only Apple bucked the trend, up 0.43% at $343.46, the sole riser on the watchlist^[1]. Index-level weakness matched the singles: the Nasdaq-100 ETF (QQQ) fell 2.50% to 709.80, the S&P 500 ETF (SPY) 1.11% to 763.25, the tech sector ETF (XLK) 3.02%, and the semiconductor ETF (SMH) 3.41% — the worst of the ETF bunch^[1]. For the full week, the Nasdaq-100 lost 5.5%, a closely watched semiconductor gauge logged its longest losing streak since 2008, and the S&P 500 fell nearly 2%, its worst week since late June^[2].

Monday in Asia: Korean stocks rebound, US futures climb

Sentiment turned reparative into Monday. South Korea's KOSPI swung from a more-than-3% drop to gains, powered by Samsung and local fund buying; the Nikkei 225 pared losses from over 2% to about 1%, while Australia's S&P/ASX 200 rose around 0.37%^[2]. US futures climbed in tandem, with Nasdaq-100 futures extending gains to 1% and S&P 500 futures up 0.5%^[2]. Kyle Rodda, senior analyst at Capital.com, argues the market is in a "typical period of high uncertainty," with AI earnings, rate uncertainty and mid-term elections all weighing on investors at once^[2].

Follow the money: Nvidia takes two-thirds

Morgan Stanley has for the first time mapped the full 2027 flow of AI capital spending: of the $1.58 trillion in hyperscaler outlays, roughly 67% of the roughly $1.03 trillion in S&P 500 revenue from AI capex — about $690 billion — flows to Nvidia alone, with the other 499 index members sharing roughly $339 billion; strip out Nvidia, and the AI capex boom contributes just 1.6% to S&P 500 sales^[3]. The report also flags that AI capex will directly drive about 25% of S&P 500 sales growth, with AI-linked companies contributing 78% of the index's 2027 earnings growth; yet hyperscalers have disclosed over $1.1 trillion in lease commitments that have not started — "synthetic capex" that sits off the cash-flow statement but is economically debt-like, making 2027 potentially the year of the widest free-cash-flow gap^[3]. Credit stress is already showing up in single names: CoreWeave, which builds data centers on heavy debt and rents them to tech firms, fell 25% last week, its worst week since November 2025, with its credit default swaps at distressed levels^[2].

The datacenter growth-inflection debate

Physical and capital constraints on the supply side are now in focus too. Barclays forecasts global (ex-China) datacenter capacity additions slowing from over 100% growth in 2026 to roughly 35% in 2027, flattening out in 2028; of some 25GW slated to come online in 2027, hyperscalers account for over 60%, while non-hyperscaler pipelines are conservatively assumed to convert at only about 40%^[4]. The report singles out CoreWeave's plan to add 5GW of datacenter capacity by 2030 as a live gauge of industry financing strain^[4]. Memory shortages are meanwhile pushing up end-user costs: Gartner says the 2026 memory crunch has already lifted PC prices by 17%^[1].

What to watch this week

Earnings are the main event: Netflix fires the starting gun on October 21, with Tesla, Meta and Lam Research reporting in a cluster on the 22nd — Meta's AI capex guidance and Tesla's delivery cadence will set the tone; on the AH side, CATL, Cambricon and Hygon report quarterly results in the same window^[5]. On the AI calendar, Nvidia's GTC Berlin opens with Jensen Huang's keynote on the 21st, after Chancellor Merz met him early on the evening of the 19th to discuss German industrial AI compute cooperation^[5]. On macro, the Financial Street Forum runs in Beijing October 19–22, the APEC finance ministers' meeting lands in Hong Kong October 20–21, and much of Europe switches to winter time on October 25, shifting data releases by an hour^[5].