The video argues Nov. 5’s AI-chip selloff was a profit-taking panic, not a business-driven break: mega-cap spenders still guided to massive 2026-27 AI capex, and the real question is whether those spending plans survive the next earnings cycle. It frames the post-election volatility as a rotation out of the biggest winners after midterm optimism and into fear around data centers, regulation, and AI job loss — but says the fundamentals still point to more buildout, not less. The actionable focus is the next set of dates and levels: Nov. 10 inflation, Nov. 17 NVDA earnings, Dec. 9 Fed, then late-January big-tech capex updates. The core call is that AI may stay volatile, but the dip in NVDA, AMD, and MU is being treated as a buying opportunity unless guidance actually rolls over.
Nov. 4’s Dow jump of 3% was the best day after a midterm in 44 years, driven mostly by oil after Democrats including Chuck Schumer said the new majority could use government funding to push for an end to the Iran war.
Cheaper oil was read as less war risk in the Straight of Hormuse, lower gas, less inflation pressure, and a friendlier Fed with less need to hike rates.
Nov. 5 flipped the narrative from oil to AI: Democrats were portrayed as hostile to data centers, the buildout was said to slow, and traders were told to sell chip stocks.
The Nov. 5 tech selloff is framed as profit-taking after a huge run, not a sudden fundamentals break; Namura’s Charlie Mccelagot called it “shooting the generals.”
Goldman Sachs had already reported in mid-October that hedge funds were cutting tech at the fastest pace in about a decade by dumping shares they already owned, so the selling did not start on Nov. 5.
MSFT reported Oct. 28 that Azure cloud revenue grew 48% to $33 billion in a single quarter, guided to about 50% growth next quarter, spent $45 billion on capex, and expects more than $50 billion next quarter and about $170 billion for all of 2026.
GOOGL reported Oct. 28 that Google cloud revenue more than doubled to almost $31 billion, kept a plan to spend about $200 billion this year, and its CFO said capex should increase significantly in 2027.
AMZN reported Oct. 29 that AWS grew 39% to $46 billion, its fastest growth in nearly five years, still expects to spend about $220 billion this year, and is projecting AWS growth of over 40% in 2027 and 2028.
META reported Oct. 21 and expects to end the year at about $140 billion of spending; the four big spenders are still on track to spend about $730 billion in 2026, maybe more, with no sign from any of them that 2027 spend is slowing.
A big chunk of that capex is still expected to flow to NVDA, AMD, and MU, and the argument is that federal policy risk is manageable because Trump can veto anything Congress passes for two years and Democrats are far short of the two-thirds needed to override.
The two bills cited as having a real shot are the Guard Act and the AI lead act, both described as bipartisan; the bigger battle is said to be at the state level, especially Texas and Ohio over power costs, community input, and local power supply for data centers.
Even in the worst-case scenario of a data-center ban, the claim is that existing data-center owners would benefit from scarcer compute, which would push more buildout into friendlier states despite red tape; the downside would be slower AI model improvement, not a kill shot.
A Washington Post/ABC/Ipsos poll in October found 57% of Americans think AI is bad for society and 2/3 oppose new data centers, a view linked to Anthropic CEO Dario Amode’s warning that AI could wipe out half of entry-level white-collar jobs.
The video says the public mostly sees AI as a fancy search engine or “autocomplete on steroids,” while users of tools like “clawed code,” ChatGPT’s codeex, and agentic AI see a much bigger gap between what AI can already do and what people think it can do.
NVDA set a record high of 243 on Oct. 6 and closed Nov. 5 around 181, about 25% lower, after losing its 200-day moving average near 205 on roughly 303 million shares, about 3x its 20-day average.
NVDA had a floor around 190 through June 29 and July 29; that floor broke on Nov. 5, making 190 the key resistance, with 171 next — the Feb. 5 and Apr. 2 floor, about 6% below the Nov. 5 close — and then about 164, the Mar. 30 low, roughly 10% below.
NVDA is set for a Nov. 17 earnings test, 12 days after the selloff; the key data points are data-center revenue, next-quarter guidance, and anything Jensen Hang says about 2027 orders, because those numbers will decide whether fear disappears or pain deepens.
AMD peaked at 658 on Oct. 6 and closed Nov. 5 around 404, down about 39%, on roughly 3x its 20-day average volume; it ended right near its 200-day around 415, broke the July 29 floor at 424, and now needs to reclaim 424 first, with 339 — the May 4 low where the big run started — the next downside level, about 16% below the close.
MU more than doubled from about 510 in early May to 1,255 on June 25, then closed Nov. 5 around 656, down about 48%; it is below its 200-day near 755, broke the July 29 floor at 738 on almost 3x normal volume, and the next floor is 510, about 22% below the close.
The setup going forward is volatile regardless: since 1950, the market has been higher 12 months after a midterm 19 out of 19 times, yet the average midterm year still sees about a 16% peak-to-bottom drop, with 2022 cited as a 25% S&P decline into October before a 14% gain 12 months later.
The watch list is Nov. 10 inflation at 8:30 ET, Nov. 17 NVDA earnings, Dec. 9 Fed with inflation data the next morning, and late-January earnings from MSFT, GOOGL, AMZN, and META for 2027 AI-spend guidance; the claim is that if spending holds up, the fear fades, and if it does not, the pain is severe.