- PSIX delivered an exceptional Q3 beat-and-raise, surpassing record Q2 2025; yet it trades at a fraction of the valuation when shares topped $100.
- If announced expansions materialize, behind-the-meter capacity could rise from ~30 GW in 2027 to >100 GW by 2030, creating excess-supply risk for incremental providers.
- Without major political change or much higher AI monetisation, the author says rapid U.S. data-center buildout is neither economically nor politically supportable.
- After reviewing ANT’s S-1 and election results, the author closed most PSIX longs, retained an OTM call, and kept sector shorts.
Investment pitches
filter:
71 74 - SpaceX's AI-linked valuation is under review: shares fell 9.1% to $123.62 on Nov. 5, while five-year CDS reached ~284bp; Bloomberg questions whether future cash generation supports the valuation.
- Reported 2026 Jan–Sep operating cash flow of ~$13.1bn versus ~$48.1bn capex implies ~$35bn simple FCF deficit; a proposed $40bn GPU finance lease adds obligations, while Customer B contributed 34.5% of Q3 revenue.
- The long-term case is cross-business compounding: Starlink, Starship, Starshield, mobile, AI and cloud may share infrastructure, but lower launch costs and other synergies remain unproven.
- Author keeps a 5–10-year thesis, citing ~$5.2bn Connectivity operating profit and ~$89.9bn cash/securities at September-end; no rush to add. AI margins, customer concentration, Starship reuse, dilution/debt, and Morgan Stanley's 2031 FCF-breakeven forecast are key tests.
70 - ASTS’s proposed reusable 20-ton Blacklift would reduce Falcon 9 dependence and complete vertical integration; first flights target late 2028 with up to five BlueBirds.
- The Firefly deal costs about $1.3B, including ~$900M cash and ~3% dilution, while near-term launch relief depends on Vulcan and New Glenn returning.
- Clear Street reiterates Buy and a $115 target, valuing ASTS at 20x 2029E adjusted EBITDA of ~$2.1B versus roughly 6x at $30.
- Pending FCC approval for Ligado’s 80-plus-year, up-to-45 MHz rights could imply author-estimated spectrum value of $16.8B versus an ~$11.7B market cap; IRDM opposes.
69 - STX's key opportunity is demand from Bitcoin Bonds: participants lock STX worth 5% of bonded BTC; the first three bonds paired ~24M STX, or 1.3% of circulating supply.
- Miners spend BTC for Stacks block production, with that BTC funding rewards; direct bonds target 3% annualized BTC yield while keeping BTC under holders’ own keys.
- Proposed PoX-6 could open staking, while borrowing against directly bonded BTC could expand utility; StackingDAO TVL reached $74.5M and Zest active loans ~$25M, though stBTC’s share is unknown.
- The author feels good holding STX into 2027, conditional on renewals, rising locked STX, and incentive-free activity; risks include ~52.6M annual STX issuance, bond unlocks, concentration, and a lower PoX-6 STX requirement.
82 - Author sees ASTS’s Ligado spectrum rights as the overlooked asset: FCC approval for up to 45 MHz across 340M POPs implies up to $16.8B, above the ~$11.7B market cap.
- The opportunity depends on FCC approval; Iridium opposes the arrangement over alleged spectrum hoarding, while Ligado disputes that claim.
- Blacklift would reduce Falcon 9 dependence but costs ~$1.3B; the Firefly deal requires ~$900M cash and ~3% dilution versus $3.7B pro forma cash.
- Clear Street reiterates Buy and $115, valuing ASTS at 20x 2029E adjusted EBITDA; current shares trade at ~6x that $2.1B estimate, with a $110 DCF cross-check.
70 - Author’s SOTP values IAM’s pipeline, partnerships, and AI platform at ~$530M bear / $1.05B base / $2.3B bull versus ~$484M market cap; base implies ~115% upside, with shares ~$1.60 above ~$385–409M pro forma cash.
- IAM1363 is the key drug upside: author assigns $30M/$250M/$500M bear/base/bull values, citing Tukysa’s $463M 2025 sales and higher Phase 1 approval rates for biomarker-selected drugs.
- The thesis needs only one approval, another large deal, or another model jump, but all three drugs remain pre-approval/early-stage; KIF18A has 10+ rivals, and SOTP depends on the author’s chosen multiples and probabilities.
70 - ASTS combines carrier alignment with the first reported ground-to-satellite handoff; the author argues SpaceX lacks comparable U.S. carrier cooperation and technical proof.
- A formal AST Defense unit, 3,900+ patent claims, and airborne, maritime, and ground applications expand the author’s government-revenue and defense-market thesis.
- Blacklift targets five BlueBirds per launch, using Firefly’s production-ready Miranda engine; the ~$1.3B commitment addresses long-term launch capacity, with first flight targeted for late 2028.
- The author remains long despite $4B+ capital needs, production at 1.7–2 versus a six-satellite monthly target, and near-term dependence on SpaceX and other launch providers.
79 - AAOI’s AI data-center revenue rose 254% year over year to $155.6 million; Q3 produced its first recent GAAP profit, $6.3 million.
- Management guides to $385–420 million of Q4 revenue and $46–60 million of adjusted profit, but sustaining that pace requires factory expansion and continued AI demand.
- The turnaround consumed $469 million in trailing-twelve-month capex, about $601 million of negative free cash flow, and substantial dilution; share count approaches 96 million.
- The author calls AAOI a high-risk turnaround, not fundamentally solid: a conservative entry is $33, while positive free cash flow, another GAAP profit, and stabilizing share count would change the view.
64 - AI token spending grew only 1.5% across August–September versus 42% in July, despite rising volumes; price competition and customer caution may cap provider revenue.
- The author argues “tokenmaxxing” is ending as enterprises confront weaker cost-benefit economics, reducing the expected pace of AI spending and data-center demand.
- The author now recommends materially reducing AI-trade exposure, including picks-and-shovels providers; high valuations require blockbuster beats while politics or a hyperscaler pullback could worsen the slowdown.
75 - MRVL closed at $133.55, about 60% below its high and ~16x Morningstar’s FY2028 adjusted-earnings forecast; the author prefers $112–120, or $100 conservatively, versus Morningstar’s $326 fair value.
- Custom AI chips and optical links drive the thesis: management targets custom-chip revenue doubling in FY2028 and again to ~$10bn in FY2029; Morningstar expects optics to grow ~60% annually, alongside a $70–90bn FY2031 revenue target versus $8.2bn in 2025.
- AI spending is the key dependency; customer concentration, larger rivals, geopolitics, and dilution add risk. Morningstar calls the all-stock Lumentum purchase overpriced, cutting fair value from $360 to $326.