Thursday’s selloff hit GOOGL hardest in its history: down almost 13%, with almost half a trillion dollars erased in one session; Friday the Nasdaq fell about 3%, and the S&P 500 closed at 7,580, roughly 3% below its record and back below its 50-day while still almost 4% above its 200-day.
Since mid-September, the S&P 500 has been boxed between about 7,550 and 7,820; the video ties that range to a weak short-term trend but says 5% pullbacks are normal almost every year and 10% corrections show up about once every 1.8 years.
History is used to argue against waiting for the perfect entry: from 1928, after a 10% S&P drop, only 4 of 10 cases went on to a 20% decline; in the dot-com top example, $10,000 all at once took almost 7 years to recover, while $100 a month got back near break-even in about 3 years; in 2008, all-in took more than 5 years, while the monthly plan broke even by November 2009.
Midterm-year seasonality is the core macro trigger: since 1950, the S&P 500 was higher 12 months after election day in all 19 midterm cycles, averaging close to 15%; from September 30 of the midterm year to June 30 of the next year, it was up 19 of 19 times with an average gain of about 20%.
The political setup is framed as stability: Democrats look likely to take the House and perhaps the Senate, Trump still keeps veto power, and the base case is 2 years of stability; markets are said to prefer that to another tariff cycle because Trump’s tariff power would be constrained in that mix.
The capex scare thesis is rejected with GOOG and MSFT data: GOOGL spent $51 billion on capex in one quarter, more than double a year ago, while free cash flow was negative $9 billion and management said next year spending will 'increase significantly'; the counterpoint is that cloud revenue is delayed, not destroyed.
GOOGL cloud revenue is up 103%, after prior growth rates of 63% and 82%, with cloud margin rising from about 24% to about 36% and cloud profit roughly tripling; MSFT Azure grew 48% faster than management guided, and MSFT capex missed because 'some third-party capacity deliveries' expected in Q1 were delayed, which is read as a capacity shortage, not weaker demand.
INTC fell about 31% Friday because PC customers are struggling with memory prices and availability; this month Intel’s CEO said the AI boom will keep memory short for 2 to 3 years, reinforcing the idea that higher capex can be a supply bottleneck rather than proof businesses can’t make money.
The sentiment backdrop is contrarian: AAII bearishness is 41% versus a long-term average near 31%, Imperva says more than half of internet traffic is bots, and the video argues bots plus fear content amplify red-day doom, making bullish calls look naïve while bears are seen as smarter.
UCTT is pitched as a beaten-down semiconductor equipment supplier: it closed Friday at $49.71, about 66% below its roughly $144 high; last quarter revenue hit a record $759 million, up 49%, and next-quarter guidance is above Wall Street.
UCTT’s business is tied to fab equipment makers: it builds ultra-pure gas lines, valves, and subsystems used in chipmaking equipment, with LRCX and AMAT described as the two biggest order drivers; the October 22 drop is blamed on LRCX saying its guide would have been higher without near-term supply-chain constraints, implying UCTT is the supplier bottleneck.
UCTT trades around 9 to 10 times next year’s expected earnings with gross margin near 17%; the two biggest customers are 61% of revenue and the company has been issuing new shares, but catalysts include reaching the 800 to 860 million guide and any 2027 spend update from LRCX and AMAT.
UCTT’s setup is defined with concrete levels: since 2008 it has had 6 years down more than 30%, then was up the next year in 5 of those 6 cases and up more than 80% in 4 of them; after the July 29 low it rallied about 37% in 2 weeks, with 44 to 46 as the floor, 69.5 as overhead resistance, 57 first upside, then 61 to 64, then 69 to 70, and a break below 44 as the invalidation.
CRDO is the AI-interconnect play: inside data centers, servers and switches connect with active electrical copper cables plus a smart chip; management expects more than $600 million this fiscal year, revenue was $479 million up 115%, and this was the seventh straight quarter of triple-digit growth.
CRDO closed Friday at $150.99, down about 51% from the roughly $309 June high; the stock trades around 24 times this year’s earnings with about $700 million of net cash, while the average analyst target is about $284, almost a double from current levels.
CRDO’s risk is that growth is slowing from 272% three quarters ago to 115% now and the business is concentrated, but the next early-December report and optical-ramp updates are framed as major catalysts; the chart trigger is a close below $143 after three tests in six trading days, with upside targets at 175 and then 194 to 202.
MRVL is pitched as a second AI-networking and custom-chip beneficiary: about 79% of revenue is data centering, it helps build custom AI chips and data-movement chips for hyperscalers that do not want to rely only on NVDA, and year-over-year growth has moved from 22% to 28% to 37% with this quarter guided at 52%.
MRVL fell 13% Thursday and another 10% Friday, closing at $196.40, about 40% below its $330 June high; at its October investor day it raised next year’s revenue target to about $20 billion, but the selloff is tied to hyperscaler capex fear even as spending is described as delayed, not canceled.
MRVL has about $1.7 billion of free cash flow over the last year and is buying back stock, but it also carries about $5 billion of long-term debt and depends heavily on a few giant custom-chip customers; the early-December report versus a $3.15 billion guide and the custom-silicon ramp are the key catalysts.
MRVL’s chart damage is specific: it bottomed around 200 to 201 in early September, then broke down through that level to about 190 on Friday with heavy volume of about 42 million shares on both Thursday and Friday, almost double normal; a close back above 200 repairs it, with upside targets at 218 to 223 and then 245 to 252.
The backtest argument on MRVL is that the market repeatedly declares it dead too early: in May 2025 it was down about 45% on the year around $55 and called a former AI favorite that would never wake up, yet about 13 months later it hit 330; this year it went from about 158 at the end of April to 330 by June 18.
GOOGL is the centerpiece of the selloff and the steadiest name on the list: Friday it closed at $279.84, down about 32% from the roughly $408 May high, yet quarter metrics still showed cloud revenue up 103%, total revenue up 25%, operating income up 30%, Search up 12%, and Gemini past a billion users.
GOOGL now trades at about 14.6 times trailing earnings, the lowest since 2012, when it was just a search company without cloud or AI; today cloud is doubling and producing profit, so the market is paying a 2012 price for a far more diversified business.
GOOGL’s sell case is that free cash flow is negative because of heavy AI investment and changing search behavior, but the counter is that the spending builds a dominant position in a lucrative future market; TPU deals, Gemini, and search volume across platforms are cited as the reasons the stock is a discount.
GOOGL’s history is used as the strongest buy-the-dip case: it has only had two years down more than 30%—2008 at -55% and 2022 at -39%—and the following years were up 93% and 57%; after the last capex scare it bottomed around 278 and hit about 408 by May 18, a roughly 47% gain in about 7 weeks.
The conclusion is explicit: GOOGL is called a great buy, and missing the dip is framed as a regret; the full four-stock list is UCTT, CRDO, MRVL, and GOOGL.
The AAII survey, Imperva bot-traffic stat, and the reliance on fear-driven engagement all feed the same trigger: sentiment is more bearish than normal while fundamentals in the selected names are still accelerating, which the video treats as a setup for a violent snapback rather than a regime break.