#156 Microsoft, Meta, Robinhood, KI-Aktien, Südkorea, FED

Summary

Theses (episode spine)

• The Fed under new chair Kevin Wash held rates in a 9-3 vote; both hosts agree there will be no further rate hike this year despite markets pricing roughly a 30% chance of one, and both read Wash as deliberately vague about which data the Fed is weighting, with Stefan arguing Wash is politically waiting for an Iran-war resolution / lower oil prices before he can justify a future cut.
• Microsoft’s Q2 beat across the board (Azure +43% YoY, first time crossing $100B annualized run-rate, 30M paid Copilot seats, positive free cash flow, capex below expectations) drove roughly a 17% single-day stock pop, one of the largest in the stock’s history; both hosts are personally invested and in profit.
• Meta’s Q2 beat on revenue (+28% YoY) but missed EPS (-13% YoY) and raised full-year capex guidance to $130-145B, sending the stock down roughly 9% the same day; both hosts are personally invested and in profit despite the drop.
• Max’s long-term thesis for Meta is a proprietary first-party data moat (WhatsApp, Instagram, Facebook) that becomes more valuable as enterprise AI shifts from public-internet data to owned data; he sees Microsoft’s edge instead as a more resilient enterprise-software business model.
• Stefan is skeptical of Meta specifically, arguing it doesn’t yet monetize its data with AI and citing Zuckerberg’s history of capital-burning (Metaverse, VR/Ray-Ban glasses); despite citing bullish Wall Street ratings for both stocks, he says he personally “would not buy either” Microsoft or Meta at current levels.
• Both hosts frame the roughly -41% Kospi drawdown since late June as a leveraged-retail-investor liquidation event (nine circuit-breaker halts this year alone, more than half of the index’s 26-year total) rather than a fundamentals problem: SK Hynix and Samsung both posted extraordinary Q2 results even as their stocks and the broader AI trade got crushed.
• Max calls SK Hynix a “no-brainer” on valuation grounds — a forward P/E around 4-5x given continuing DRAM/HBM price hikes (+30% QoQ average selling price) and multi-year long-term supply agreements that lock in demand into 2028-29.
• Both hosts flag circular-financing concerns (AMD-Core Scientific, Nvidia’s $50B H aid leasing deal) and open-source-model credibility issues (Kimi K2, DeepSeek) as reasons for near-term caution on the broader AI trade — 69% of S&P 500 IT names are down more than 20% from highs, while Apple, which does not build its own AI infrastructure, has held up.
• Max says he is personally waiting for a new “application ROI” catalyst before broadly adding to AI names, preferring companies that have already proven infrastructure-driven revenue growth.
• On Robinhood, Stefan is emphatically bullish and bought additional long-dated call options after the post-earnings dip, arguing the shrinking share of crypto in total revenue makes the business healthier; Max, also long, agrees and expects the stock “significantly higher” in 6-12 months.

$MSFT (Microsoft)

Convergence / divergence: Both hosts are already invested and in profit, and both praise Microsoft’s Q2 results, but only Max frames it as an active buy; Stefan treats the consensus $559 target and 42-buy/6-hold/0-sell rating spread as an external data point rather than his own recommendation, and explicitly says he would not add here.

Speaker calls:

• Max (bullish, long-term): Sees Microsoft’s Azure growth (+43% YoY, first quarter over $100B annualized) and 30M paid Copilot seats as evidence of a durable, accelerating cloud/AI business, backed by a business model he views as more resistant to enterprise budget cuts than Meta’s.
• Stefan (neutral): Recited Wall Street’s consensus target of $559 versus the $456 price at recording (about 22.5% upside, 42 buy / 6 hold / 0 sell ratings) but said he personally would not buy Microsoft here, questioning whether Copilot AI revenue can be cleanly separated from bundled Office/Cloud subscription revenue.

Cross-check:

• Price: $479.20 (P/E 27.45 trailing / 25.05 forward, market cap ~$3.6T). Next earnings: ~late October 2026 (est.; last reported 2026-07-29).
• Recent headlines worth knowing: Shares have kept climbing since the July 29 beat, now above the $456 level cited on the show.
• Inconsistencies: none flagged — price has moved toward, not away from, the hosts’ bullish framing, though the gap to the $559 consensus target has narrowed since recording.

$META (Meta Platforms)

Convergence / divergence: Both hosts are already invested and in profit, but they diverge sharply on going forward: Max sees a durable data moat and expects Meta to keep growing in the AI era (while acknowledging it would likely fall harder than Microsoft in a downturn); Stefan is openly skeptical Meta can monetize its AI investment and says flatly he would not buy either Meta or Microsoft at current levels.

Speaker calls:

• Max (bullish, long-term): Argues Meta’s edge is its proprietary first-party data (WhatsApp, Instagram, Facebook), which becomes more valuable as enterprise AI shifts from public-internet data toward owned data that competitors like Claude cannot access.
• Stefan (bearish): Cited the Wall Street consensus target of about $797 (roughly 49% upside from $535, 40 buy / 8 hold / 0 sell ratings) but said he would not personally buy Meta, citing its unclear AI monetization path and Zuckerberg’s track record of spending heavily on the Metaverse and VR/Ray-Ban glasses without a clear return.

Cross-check:

• Price: $556.71 as of 2026-07-31 (P/E 20.98 trailing / 17.28 forward, market cap ~$1.42T). Next earnings: ~late October 2026 (est.; last reported 2026-07-29).
• Recent headlines worth knowing: Shares have already rallied back above the $535 level cited on the show; average analyst target is now cited at $769 (+38%), still Strong Buy consensus, with some outlets flagging the 17x forward P/E as cheap relative to Meta’s 5-year average.
• Inconsistencies: none flagged — the post-earnings rebound has partly undercut Stefan’s near-term caution, though his underlying monetization skepticism is a separate, still-open question the transcript doesn’t resolve.

$HOOD (Robinhood Markets)

Convergence / divergence: Both hosts converge on bullishness and both hold positions, with Stefan actively adding via long-dated call options on the dip. Both agree the declining share of crypto revenue is structurally healthy for the business rather than a warning sign, differing only in that Stefan quantifies his conviction with specific analyst price targets while Max keeps his call more qualitative.

Speaker calls:

• Stefan (bullish, 6-12 months): Bought more long-dated call options after the post-earnings dip, framing it as a buy-the-dip opportunity; flagged the EPS beat as partly a low-quality one-time gain from consolidating the Robinhood Ventures Fund I, but argued crypto’s shrinking share of transaction revenue (about 1/8, versus a larger $156M from event contracts/prediction markets) makes the underlying business healthier, with a possible tailwind from a seasonal Bitcoin cycle rally expected around October.
• Max (bullish, 6-12 months): Also long and in profit; agrees the declining correlation to Bitcoin/crypto is a healthy structural shift as equities, options, Gold subscriptions and Trump Accounts grow, and expects the stock to be significantly higher in 6-12 months.

Cross-check:

• Price: $90.97-$92.15 as of 2026-08-03 (P/E ~38-40 trailing, market cap ~$82.5B). Next earnings: 2026-11-04.
• Recent headlines worth knowing: Stock has risen from the ~$87-88 level discussed on the show toward $91-92, consistent with the hosts’ dip-buy thesis; broader analyst coverage shows 22 Buy vs. 2 Sell ratings.
• Inconsistencies: hosts characterized the bank ratings they reviewed as having no sell ratings; broader coverage does include 2 sell ratings, so their sample wasn’t fully representative of all analyst coverage.

000660.KS (SK Hynix)

Convergence / divergence: Stefan did not give a company-specific call on SK Hynix — his commentary was limited to the broader Kospi/leveraged-retail-investor dynamic, so only Max’s valuation thesis is represented here.

Speaker calls:

• Max (bullish, long-term): Calls SK Hynix a “no-brainer” after the Kospi drawdown, citing Q2 results (revenue +257% YoY, operating profit +557% YoY, record 76% operating margin, EPS +1,242% YoY) alongside a forward P/E around 4-5x, driven by 30% QoQ DRAM/HBM price hikes and multi-year long-term supply agreements that secure demand into 2028-29.

Cross-check:

• Valuation: trailing P/E 16.5, forward P/E roughly 4.4-5.9x, market cap ~KRW 1,282 trillion. Next earnings: next quarterly report expected ~late October 2026 (last reported 2026-07-29).
• Recent headlines worth knowing: Stock up over 580% over the trailing 52 weeks despite the Kospi-wide drawdown discussed in the episode.
• Inconsistencies: none flagged — independently sourced forward P/E figures (4.4-5.9x) closely corroborate Max’s on-air estimate of 4-5x.

Topics discussed

FOMC decision under new Fed chair Kevin Wash

Summary: The Fed held rates in a 9-3 vote (three members favored a hike) under new chair Kevin Wash. Markets whipsawed intraday — the S&P 500 fell, then rallied about $1T, then dropped roughly $1.1T in 30 minutes after Wash’s press conference, partly on a new 30-year Treasury yield high of 5.2%. Wash said “the bond market has done the work for us” and declined to specify which economic indicators the Fed is weighting.

Speaker views:

• Max: Believes no further hike is likely this year; notes seven of twelve members would need to vote for a hike to overrule the chair, and Wash is likely waiting for his internal task-force review before committing to a policy direction.
• Stefan: Says flatly there will be no hike this year, calling Wash evasive and deliberately withholding forward guidance; believes Wash is politically waiting for an Iran-war resolution and falling oil prices before he can justify future rate cuts, and that the U.S. cannot afford higher rates given war spending and debt-service costs already exceeding the defense budget.

Potential impact: Both hosts think the Fed will remain a secondary market driver until the next meeting on September 16, absent a sharp escalation in Middle East oil-supply disruption pushing oil toward $120-150.

Kospi/South Korea market crash despite strong chip earnings

Summary: The Kospi fell about 41% from its late-June peak (about 40 trading days), triggering nine circuit-breaker halts this year alone (versus 15 total in the index’s 26-year history), even though SK Hynix and Samsung both reported extraordinary Q2 results. The hosts attribute the crash to leveraged-ETF retail investors being forced out of the market rather than to fundamentals, and note foreign investors face limited direct access to Korean equities (ADRs only).

Speaker views:

• Max: Sees the crash as disconnected from fundamentals given SK Hynix’s and Samsung’s record margins and profit growth, and frames the resulting valuation drop as an opportunity.
• Stefan: Compares it to Korea’s earlier leveraged silver-trading crash, calling it the same speculative pattern repeated in AI-linked stocks; notes Wall Street increasingly focuses on which downstream companies actually benefit from implementing AI (e.g., cost/headcount savings).

Broader AI-trade correction and circular-financing concerns

Summary: 69% of S&P 500 IT-sector names are down more than 20% from recent highs, meeting the bear-market threshold; individual AI-infrastructure names fell 50-70% from highs (Coreweave, Oracle, Sandisk, SK Hynix, Corning, Nebius, Marvell, ARM, Astera Labs), while Apple — which does not build its own AI infrastructure and licenses AI (e.g., from Alphabet for Siri) instead — held up. Hosts also flagged new AMD-Core Scientific and Nvidia-H aid ($50B leasing) deals as raising circular-financing concerns, and criticized Chinese open models Kimi K2 and DeepSeek for reliability and credibility issues, including a leaked DeepSeek investor transcript acknowledging continued reliance on expensive Nvidia chips.

Speaker views:

• Max: Says hedge funds drove much of the selling by rapidly cutting U.S. tech exposure; believes the market needs a new “application ROI” catalyst (evidence AI spend is translating into monetizable products) before broadly re-entering AI names, and prefers companies that have already proven infrastructure-driven revenue growth.
• Stefan: Questions whether cheaper Chinese AI models are credible, citing Kimi K2’s poor translation quality, capacity limits for paying users, and instances of the model misidentifying itself as ChatGPT or Claude; sees this as evidence the “cheap Chinese AI” narrative is overstated.

Potential impact: Max expects that once a clear application-ROI catalyst appears, momentum in AI-infrastructure names could resume, but sees risk of a further 5-10% market pullback in August/September (historically the two weakest months) before any resumption.