InvestAnswers: AI trade roars back as $63B TPU deal eases fear-driven selloff

James says last week's AI panic reversed on a wave of capital commitments

James opens by framing the prior week's selloff as driven by fears that "AI is out of money," inflation worries, and expectations the Fed would hike (it did not). He argues that narrative reversed once Bank of America, Blackstone, Citi and Morgan Stanley "cut a check for $63 billion to buy some Broadcom TPUs," while SpaceX also raised $40 billion at favorable rates. He says Neocloud and memory-related fears "partly reversed... big time," pointing to Micron and SK Hynix as examples, though he notes oil remains "still a problem" above $110 due to strikes.

He runs through his weekly scorecard, admitting two misses: Solana's $104 support, which failed to $103, and a call that MicroStrategy would bounce back, which didn't happen after comments from Chamath Palihapitiya (saying Bitcoin was going to zero) and a conflicting call from "Jason" targeting $50K. He tallies "60% pure targets and with partials 89%" accuracy for the week.

Bitcoin, Solana and Ethereum ETFs diverge sharply

James says Bitcoin broke his $81,600 support but recovered quickly, and is "back above that 816 support," with resistance at roughly $87K. He expects "November and December to be pretty okay for Bitcoin" despite a weak October. On the Bitcoin-gold ratio, he maintains his view that it will move "from what was about 12 ounces of gold per Bitcoin" back toward "40," pending Bitcoin reclaiming $82K.

He flags a divergence in institutional flows: "Fidelity and BlackRock bought zero, zero Ethereum ETFs for the last month. They only sold," while Bitcoin ETF flows resumed buying over the last two days. Solana broke his $104 support level, trading at "102.15" during the show. He has no firm thesis on why Ethereum is underperforming, saying only "maybe they know what we know."

Tesla and SpaceX rally, Google tanks

James highlights Tesla's run from roughly $345 to $405 over seven trading days after clearing $380, calling it a "clear buy signal" with no sell signals, and floats targets of $450 or even a new all-time high near $500 "if the ramp of the cyber cab continues." He describes SpaceX as his preferred long-term holding, citing forecasts of "10 to 13 gigawatts of compute" and revenue potentially reaching "350 maybe 400 billion" next year against a market cap "under two trillion." He says he has been "buying SpaceX pretty much every week for the last 14, 15 weeks."

Google, by contrast, is "deep under the 200 day moving average" by about 15%, the widest gap in roughly five years outside the "tariff tantrums" period. James attributes the drop to a miss on earnings per share ($2.12 versus an expected $3) and a capex guidance increase to "195 to 205 billion," which "Wall Street hates." He says Google is now his "smallest position" among this group "for a reason," though he still sees a long-term price target "north of 400."

Chip and memory names seen as the stronger AI trade

"If you're in Marvell, Micron, SK Hynix, Astera Labs, Broadcom, yeah, it's back. If you're in Neoclouds, you're going to get stuffed." — James

James runs through several chip-adjacent names he says bounced exactly off levels he called the prior week: SK Hynix off "1.182 million won," Marvell off its rising channel, Astera Labs off "$16," Broadcom off the "330 kill zone," and Lam Research off "level three." He ties Lam Research's strength partly to work with Tesla and SpaceX on "Terrafab." He also notes Credo bounced off its "130s" support and calls it part of the verified AI trade recovery.

On Baidu, he notes the stock is "at the same price it was 17 years ago" and calls the current level "generational support," but says any recovery to $160 depends on revenue growth returning, which "might take six months or a year."

Views on MicroStrategy, Strive and portfolio strategy

Asked whether to sell MicroStrategy at a loss to buy SpaceX or Tesla, James says SpaceX is "the better value here right now" and, if holding for one to three years, would prefer SpaceX, but suggests layering: taking profit on MicroStrategy if it rallies to "160, 175" on a Bitcoin spike and reallocating into SpaceX. He separately flags Strive as buying "3,4,000 Bitcoin a week" and running at a "much higher NAV premium than MicroStrategy," while warning that Bitcoin treasury companies offering "12, 13, 14% dividends on preferreds" are not sustainable "unless... they can do it with a positive NAV premium."