#145 Take Two, SpaceX, AMD, Nvidia, Intel, EchoStar, Elon Musk, Starlink

Summary

Theses (episode spine)

• Stefan has invested €54,000 in Take-Two Interactive call options (three tranches) targeting €100,000 by mid-November 2026, betting GTA 6 releases on 19 Nov 2026; at recording (18 May) the option was +30% (€15,000 profit).
• The Take-Two Q1 earnings on 21 May are irrelevant to Stefan’s thesis — the only catalysts are: (A) official GTA 6 release date confirmation, and (B) revenue projection for GTA 6. Stefan plans to sell before the game launches to avoid post-launch sell-the-news.
• AMD Q1 2026 results confirmed the agentic AI thesis: data-centre revenue $5.8B (+57% YoY), Q2 guidance +70% YoY, net profit $1.4B (doubled YoY), FCF $2.6B beat $1.7B estimate. AMD-Meta deal up to 6 GW of Instinct GPUs.
• The agentic-AI (CPU-intensive) trend is the primary driver for AMD and Intel: agentic workloads require more CPUs relative to GPUs — AMD Server CPU TAM revised upward from $60B to $120B by 2030.
• Stefan holds 1,000 AMD shares (€365,000 at recording, ~+18,000% from entry in 2011 at below €2) plus a leveraged AMD call option currently +70%. Max holds AMD at ~+230%.
• EchoStar (SATS) is Stefan’s SpaceX IPO proxy: EchoStar sold spectrum licences to SpaceX and received SpaceX shares; at recording, EchoStar’s SpaceX stake could equal its entire market cap at a $2T SpaceX valuation.
• Stefan bought EchoStar in January 2026 (via options at ~€8K); Max bought shares at ~$109 and was +26% at recording ($136). Both plan to sell before or shortly after the SpaceX IPO.
• SpaceX IPO expected mid-June 2026 (prospectus possibly 20 May); Elon Musk reportedly wants CEO-for-life status. Max warns: $2T valuation for an unprofitable company is “jenseits von gut und böse” (beyond all reason); post-IPO sell-off risk is real.
• Intel Q1 also strong; same agentic AI CPU demand tailwind as AMD. Intel grows primarily via price (ASP); AMD grows via unit market share gain (+35% QoQ units). Computex (2–5 June) may bring new Intel foundry announcements.
• Hardware price inflation (SSDs, HDDs, RAM up 3–5x in under a year) is a structural tailwind for AMD and Intel as AI data-centre buildout accelerates.

$TTWO (Take-Two Interactive)

Convergence / divergence: Stefan is aggressively long via options and has built conviction over nearly 4 years of rolling GTA 6 derivatives. Max agrees the thesis is sound and the Q1 earnings are irrelevant to the GTA 6 story but has not committed capital at the time of recording.

Speaker calls:

• Stefan (bullish, medium-term, targeting >$300): Has been invested in TTWO derivatives since 2022. Q1 earnings on 21 May are “irrelevant” — the market only cares about (A) GTA 6 release date confirmation and (B) revenue projections. Dismisses three common bearish arguments: “already priced in” (stock is 10% below ATH of $264, not at $300+), “will be delayed again” (CEO confirmed launch and summer marketing campaign), “will be woke/bad” (irrelevant since Stefan plans to sell before launch). GTA franchise has generated more revenue than any film franchise including Marvel, Star Wars, Disney. Each future GTA 6 trailer will trigger additional stock push. Plans to sell before 19 Nov to capture the pre-launch run and avoid sell-the-news risk.
• Max (neutral, watching): Agrees Q1 earnings are beside the point. Draws Cyberpunk 2077 analogy — the game can be great yet the stock may disappoint post-launch if expectations are too high. Notes buying before earnings is normally gambling but here the game-launch catalyst is the unique driver. Had not committed to a position at time of recording.

Cross-check:

• Price: $227.70 (P/E N/A — unprofitable, market cap $44.1B). Next earnings: 2026-08-10.
• Recent headlines worth knowing: 21 May Q4 2026 earnings confirmed GTA 6 for 19 Nov 2026; summer 2026 global marketing campaign confirmed; FY2027 Net Bookings guidance $8.0–8.2B (up from $6.72B); EPS $0.80 beat $0.58 estimate. Stock rallied post-earnings. TD Cowen projects >40M units sold in first 12 months.
• Inconsistencies: Stefan’s thesis was fully validated on 21 May (3 days after this episode). GTA 6 date confirmed, revenue guidance given. Stock at $227.70 is below the $242 recording price, meaning the option P&L path to €100K depends on whether the stock breaks above $264 ATH before November.

$AMD (Advanced Micro Devices)

Convergence / divergence: Both speakers are long-term AMD bulls who have held through the entire AI cycle. Both would wait for a pullback before adding significantly more given the strong recent run (+114% YTD at recording, +265% 1-year USD).

Speaker calls:

• Stefan (bullish, long-term): AMD is his largest position (~39% of portfolio). Q1 results: revenue $10.3B (+38% YoY), data-centre $5.8B (+57%), net profit $1.4B (doubled YoY), FCF $2.6B (+53% vs. $1.7B estimated). Q2 data-centre guidance +70% YoY. Agentic AI driving CPU demand inflection — “instead of 1 CPU and 8 GPUs, you now need 4 CPUs and fewer GPUs.” AMD grows via units (+35% QoQ) vs Intel which grows via price. Geopolitical risk: all leading-edge AMD chips produced at TSMC in Taiwan; China-Taiwan conflict would be the worst scenario. Will hold but wait for a pullback before adding.
• Max (bullish, long-term): Server CPU TAM doubled to $120B by 2030 (per AMD management). Free cash flow $2.6B beat $1.7B — AMD is a cash machine. Gross margin 55%, FCF margin 25%. AMD-Meta deal (up to 6 GW Instinct GPUs, MI450 series) is the company’s largest-ever deal. Would wait for a pullback or post-Computex dip before adding. Also notes geopolitical risk from TSMC concentration.

Cross-check:

• Price: ~$456–481 range (24 May 2026; all-time high $481.37 on 22 May). P/E ~50–55 TTM. Market cap ~$685–762B. Next earnings: 2026-08-04.
• Recent headlines worth knowing: All-time high reached 22 May 2026 at $481; stock +114% YTD, +265% 1-year (USD). Q1 data-centre revenue +57% confirmed; Q2 guidance +70% reiterated. AMD-Meta and AMD-OpenAI deals confirmed (both MI450 series). Lisa Su’s server CPU TAM estimate doubled.
• Inconsistencies: Stefan said his ~1,000 shares were worth €365,000 at recording (€365/share = ~$390); at the current ~$456–481 level, the position has grown materially further since the episode — Stefan’s thesis is playing out ahead of expectations. Max’s +230% gain has also expanded.

$INTC (Intel)

Convergence / divergence: Both speakers acknowledge Intel’s dramatic recovery and the agentic AI tailwind, but neither holds a current position — Stefan sold his Intel stake when it was at ~€37 and regrets the decision.

Speaker calls:

• Stefan (bullish, watching): Intel Q1 also strong on the same agentic AI / CPU demand theme as AMD. Intel grows primarily via higher average selling prices (ASP), contrasted with AMD which grows via unit market share gains. Computex (2–5 June) may bring new Intel foundry (TSMC-competing) partnership announcements. Intel’s “Crescent Island” GPUs for inference and agentic AI could also produce positive news at Computex.
• Max (cautiously bullish): The turnaround is real — stock went from €37 low (when Stefan sold) to ~€100. Agrees Computex could be a positive catalyst. Notes buying at ~$107 after a +174% 3-month run requires patience.

Cross-check:

• Price: ~$118.50 (P/E N/A on GAAP basis — still recovering; market cap $602B). Next earnings: 2026-07-23.
• Recent headlines worth knowing: Intel stock +483% 1-year, all-time high $132.75 on 11 May 2026. Strong Q1 results; agentic AI CPU demand confirmed as key driver. Computex 2–5 June expected to bring foundry news and Crescent Island GPU announcement.
• Inconsistencies: Stefan cited Intel at ~€100 at recording; it has since risen to ~$118.50 — higher in USD terms, though EUR/USD conversion matters. Stefan’s regret at selling at €37 is underscored by the subsequent move.

$SATS (EchoStar — SpaceX IPO proxy)

Convergence / divergence: Both speakers are in EchoStar as a SpaceX IPO proxy and both plan to exit before or shortly after the IPO, anticipating that EchoStar holders will rotate into SpaceX directly once it is listed.

Speaker calls:

• Stefan (bullish, short-term, sell-before-IPO): EchoStar sold spectrum licences to SpaceX and received SpaceX Class A shares in return (“SpaceX = MicroStrategy, EchoStar = MicroStrategy for SpaceX”). Estimated SpaceX stake at ~2% of SpaceX; at a $2T SpaceX valuation this would equal ~$40B — approximately EchoStar’s entire market cap at recording. SpaceX IPO prospectus expected ~20 May (two days after episode). Plans to sell leveraged options position before IPO date, not after, to avoid EchoStar holders rotating into SpaceX directly.
• Max (bullish, short-term): Holds shares at ~$109 (+26% at recording). Confirms SpaceX IPO story is “the most likely scenario” for early-to-mid June. Warns the $2T valuation is “jenseits von gut und böse” (way beyond reason) for an unprofitable company — draws analogy to Cerebras IPO (100% day-one pop, then faded to +70%). Notes Alphabet also holds a SpaceX stake but EchoStar’s impact per dollar invested would be much larger.

Cross-check:

• Price: ~$122.74–132.27 range (24 May); approximately $126.00. P/E N/A. Market cap ~$39.5B. Next catalyst: SpaceX IPO June 2026.
• Recent headlines worth knowing: SpaceX IPO prospectus filed 20 May 2026 (the day after this episode aired); IPO expected June 2026 at $1.75–2T valuation. EchoStar’s SpaceX stake: 52M Class A shares, theoretical value $11B per SEC filings (at ~$212/share basis); at current implied SpaceX valuation ~$27–31B per analysts. New Street Research Buy/$161; TD Cowen Buy/$155. Stock hit all-time high of $147 on 18 May (day of recording).
• Inconsistencies: The SpaceX prospectus was filed the very next business day (20 May), validating Stefan’s timing. However, the stock at $126 is now below the $136 recording price, suggesting some “sell the prospectus” reaction. Stefan’s 2% stake estimate is partially validated by SEC data (52M shares × ~$212/share = ~$11B at one valuation basis; analysts put it at $27–31B at higher valuations). Max’s math is directionally correct but imprecise given SpaceX’s private valuation uncertainty.

$NVDA (Nvidia — context only)

Nvidia was mentioned in the episode primarily as context: it is the dominant GPU supplier that AMD competes with in AI data centres; AMD is gaining market share with Instinct GPUs. Nvidia has no earnings on Q1 results data per the episode (Stefan mentioned Nvidia reports “this week” — around 20 May). Neither speaker expressed a primary position or investment thesis on Nvidia in this episode.

Cross-check (context):

• Price: ~$215.33. Market cap: ~$5.22–5.32T (world’s most valuable company). Next earnings: Q1 FY2027 results expected ~20 May 2026 (may have already been released at time of writing).

Topics discussed

Agentic AI — CPU Demand Inflection Point

Summary: Both speakers argue that agentic AI (autonomous multi-task AI systems, like Claude Code agents) is structurally shifting data-centre hardware demand from GPU-heavy to CPU-intensive. In traditional AI inference, the typical configuration might be 1 CPU to 8 GPUs. In agentic architectures, the CPU orchestrates multiple autonomous sub-tasks, requiring more CPU cores per GPU. Stefan uses the analogy: GPUs are kitchen helpers, CPUs are head chefs — as each helper (GPU) becomes smarter and handles more tasks independently, you need more head chefs (CPUs). AMD’s data-centre revenue +57% YoY confirms this shift is already showing in earnings. Intel Q1 also benefited from the same trend.

Speaker views:

• Stefan: Agentic AI demand for CPUs is just beginning; the shift is visible in AMD Q1 and Intel Q1; hardware price inflation (SSD, HDD, RAM up 3–5x) shows a genuine supply crunch in AI-adjacent components. AMD growing via units is structurally superior to Intel growing via price.
• Max: Server CPU TAM doubled by AMD management to $120B by 2030. AMD’s fabless model (all leading-edge production at TSMC) is a geopolitical risk if China acts on Taiwan, but for now is a competitive advantage via TSMC’s manufacturing leadership.

Potential impact: Sustained tailwind for AMD and Intel through at least H2 2026. Computex (2–5 June 2026) could be a near-term catalyst for both. The CPU demand shift, if sustained, is a meaningful headwind for pure GPU plays.

SpaceX IPO — The Mechanics and Risks

Summary: SpaceX is expected to IPO in mid-June 2026 at an approximate $1.75–2 trillion valuation. Elon Musk reportedly seeks CEO-for-life status as a condition. The episode was recorded 18 May; the prospectus was filed 20 May. Stefan and Max use EchoStar (SATS) as a proxy. The key structural question: how many shares will float, and at what price? Stefan draws an analogy to GTA 6 — both SpaceX and GTA 6 are massively hyped events where the pre-event run may be the better trade than the post-event hold.

Speaker views:

• Stefan: Buy EchoStar now, sell before or right at the IPO date confirmation. The SpaceX stake in EchoStar (estimated at ~2% of SpaceX) at a $2T valuation would theoretically match EchoStar’s entire market cap — essentially buying SpaceX for free with EchoStar’s other assets thrown in.
• Max: Bullish on the pre-IPO trade but warns the $2T valuation is extreme for an unprofitable company. Cites Cerebras as a cautionary analogy (25x oversubscribed, +100% first trade, faded to +70% on day one). Would not hold through the IPO lock-up period.

Potential impact: EchoStar (SATS) has a clear near-term catalyst path. SpaceX post-IPO is a higher-risk, valuation-stretched proposition. The IPO will be the largest in history at $1.75–2T.