Elon's Anthropic Deal, The Next AI Monopoly?, "FDA for AI" Panic, Trading the AI Boom

Summary

Theses (episode spine)

• xAI leased all of Colossus 1 (H100 GPUs, 300+ MW) to Anthropic, establishing “Elon Web Services” (EWS) as a fourth hyperscaler; Brad Gerstner estimates this generates $4–5B of incremental revenue for xAI/SpaceX in 2026 on top of mid-$20B analyst estimates.
• Anthropic grew ARR from ~$10B to ~$30B in Q1 2026, then from $30B to $44B in April alone; David Sacks argues that if this trajectory continues 18 more months, Anthropic will be the most valuable tech company in history and could approach $1T ARR by end of 2027.
• Sacks warned Anthropic’s trajectory could create “the biggest monopoly in human history,” invoking the Rockefeller / Standard Oil analogy and arguing AI safety rhetoric is being used — knowingly or not — as cover for regulatory capture that would entrench the current Anthropic/OpenAI duopoly.
• The reported “FDA for AI” executive order was dismissed as largely fake news by Sacks and Gerstner, who both said they spoke to NEC Director Kevin Hassett and he does not support a pre-approval regime; a statement from White House chief of staff Susie Wiles was cited as shooting it down.
• Brad Gerstner holds 25% of his portfolio in memory stocks — SK Hynix (~5x GAAP), Samsung (~6x), Micron (~7x) — arguing these valuations are not bubble territory and AI infrastructure demand justifies them.
• Gerstner expects SpaceX to trade at 40–50x revenue at IPO (implying ~$2T market cap), justified by Elon’s unique five-layer innovation pipeline (launch, Starlink, hyperscaling, orbital data centers, models/applications).
• Chamath Palihapitiya argued that tech leadership has failed at messaging and redistribution, earning a “D-minus trending to F,” and that the regulatory backlash is a direct consequence of wealth concentration without broad-based societal reinvestment.
• On AI ROI: Gerstner and Sacks see 200bps of S&P 500 operating margin expansion and hyper-accelerating cloud revenue (AWS +28%, Azure +39%, GCP +63%) as proof AI is delivering; Chamath counters there is “not a scintilla of evidence” attributable specifically to AI and says a reckoning is ~500 days away when enterprises must prove X spent yields Y returned.
• All hosts agreed AI salience among voters ranks 29th out of 39 issues, with cost of living and the economy at the top — Sacks argued AI’s actual effects (deflation, growth, construction boom) are popular even if the technology itself is not.
• Jason Calacanis made the case for concrete redistribution responses: IPO share allocations to all Americans via “Invest America” accounts, modest minimum wage increases, and universal healthcare — suggestions the VC voices on the panel largely declined to endorse.

$NVDA (Nvidia)

Convergence / divergence: Only Gerstner commented on NVDA directly. He framed it as a core AI infrastructure holding at a reasonable valuation alongside the rest of the memory/compute stack.

Speaker calls:

• Brad Gerstner (bullish): Nvidia is trading at 19x fully-taxed GAAP earnings — not bubble territory — and is central to the AI infrastructure buildout.

Cross-check:

• Price: $215.20 (P/E ~43.65 TTM, forward ~24.5x; mkt cap $5.23T). Next earnings: 2026-05-20.
• Recent headlines: Nvidia remains the world’s most valuable company. Earnings due May 20, 2026. Strong forward earnings growth expected.
• Inconsistency: Gerstner cited 19x GAAP earnings; the trailing P/E is ~43.65x and the forward P/E is ~24.5x. His figure appears to reference forward estimates — the trailing multiple is roughly double what he stated.

$META (Meta Platforms)

Convergence / divergence: Gerstner was the sole commenter. He viewed META as fairly valued at 17x GAAP alongside the other Mag-7 names, none of which he considers in bubble territory.

Speaker calls:

• Brad Gerstner (bullish): Meta is trading at 17x fully-taxed GAAP earnings, which Gerstner considers a fair valuation not in bubble territory.

Cross-check:

• Price: $609.63 (P/E ~22.15 TTM; mkt cap $1.55T). Next earnings: 2026-07-29.
• Recent headlines: Meta Q1 2026 EPS $7.31 (beat $6.67 est), revenue $56.31B (+33% YoY). Wall Street 12-month consensus target ~$840.
• Inconsistency: Gerstner cited 17x GAAP; current TTM P/E is ~22x. His figure likely references forward estimates or a slightly earlier data point.

$MSFT (Microsoft)

Convergence / divergence: Gerstner commented on Microsoft primarily as a datapoint on hyperscaler growth and margin expansion, not as a specific buy/sell call.

Speaker calls:

• Brad Gerstner (neutral): Microsoft Azure grew 39% with only ~3% headcount growth at Mag-5 companies over three years, indicating expanding operating margins; trades at 20x GAAP earnings.

Cross-check:

• Price: $415.12 (P/E ~24.11 TTM; mkt cap $3.08T). Next earnings: 2026-07-28.
• Recent headlines: Microsoft beat Q3 2026 on top and bottom lines; Azure +39% YoY; 20M+ paid Copilot seats; capex guidance raised to $190B (+61% YoY), including long-term DDR5 supply deal with SK Hynix.
• Inconsistency: Gerstner cited 20x GAAP; current TTM P/E is ~24x — directionally consistent, slightly understated.

$GOOGL (Alphabet / Google)

Convergence / divergence: Gerstner was bullish, citing GCP’s 63% growth as evidence. Sacks was more neutral — acknowledging Google’s coding strength as a competitive threat to Anthropic without making an explicit valuation call.

Speaker calls:

• Brad Gerstner (bullish): Google Cloud grew 63% in the quarter and Alphabet trades at 24x GAAP earnings; Gerstner sees this as fairly valued, not penalized.
• David Sacks (neutral): Google is very good at coding and won’t be asleep at the wheel in the AI coding competition, but Anthropic holds the inertia advantage.

Cross-check:

• Price: $399.49 (P/E ~30.06 TTM; mkt cap $4.81T). Next earnings: ~late July 2026 (Q2).
• Recent headlines: Alphabet Q1 2026 revenue $109.9B (beat), GCP grew 63% YoY. Stock near $400, market cap $4.81T.
• Inconsistency: Gerstner cited Google at 24x GAAP; current TTM P/E is ~30x. His figure likely references forward estimates.

$000660.KS (SK Hynix)

Convergence / divergence: Gerstner was the sole commenter and expressed the strongest conviction here — 25% portfolio allocation is a concentrated bet.

Speaker calls:

• Brad Gerstner (bullish): Holds 25% of his portfolio in SK Hynix at ~5x fully-taxed GAAP earnings, calling memory stocks a core AI infrastructure bet that is not in bubble territory.

Cross-check:

• Price: 1,686,000 KRW (record high as of May 2026; mkt cap large-cap KRX). Next earnings: ~Q2 2026.
• Recent headlines: SK Hynix Q1 2026 record revenue KRW 52.6T (+198% YoY), operating margin 72%, net profit nearly quadrupled. Entire 2026 production capacity already sold out. Signed multi-year DDR5 supply deal with Microsoft. Hit all-time high above 1.6M KRW in early May 2026.
• Inconsistency: None flagged. Gerstner’s ~5x GAAP claim is consistent with the extraordinary Q1 results. Stock is at record highs, reinforcing his thesis.

$MU (Micron Technology)

Convergence / divergence: Gerstner grouped Micron with SK Hynix and Samsung as his core memory/AI infrastructure position.

Speaker calls:

• Brad Gerstner (bullish): Micron at ~7x fully-taxed GAAP earnings is part of his memory stock thesis — cheap relative to the AI infrastructure demand cycle.

Cross-check:

• Price: data anomalous from search (raw figure $757 appears incorrect for MU; typical range ~$85-130). Forward P/E: 7.55x (GuruFocus). Next earnings: ~2026-06-23 (est).
• Recent headlines: Micron benefits from AI memory demand tailwinds alongside SK Hynix. Forward P/E of ~7.5x is consistent with Gerstner’s ~7x claim.
• Inconsistency: Cross-check price data appears anomalous; treat MU current price as unverified from this search. The forward P/E figure is consistent with Gerstner’s stated valuation.

$AMZN (Amazon / AWS)

Convergence / divergence: Gerstner referenced AWS primarily as a hyperscaler growth datapoint. No other host made an explicit AMZN call.

Speaker calls:

• Brad Gerstner (bullish): AWS is now on a $150B annualized run rate growing 28% — confirmation that AI ROI is showing up in cloud infrastructure revenue.

Cross-check:

• Price: $272.58 (P/E ~33.25 TTM; mkt cap ~$2.96T). Next earnings: 2026-07-29.
• Recent headlines: AWS $150B run rate at 28% growth — consistent with hosts’ cited figures. Amazon market cap up ~28% over the last 30 days.
• Inconsistency: None flagged. AWS run rate and growth figures cited by Gerstner are consistent with public Q1 2026 data.

$TSLA (Tesla)

Convergence / divergence: Both Chamath and Calacanis were bullish on Tesla through the lens of an eventual Tesla-SpaceX merger. Neither gave a price target.

Speaker calls:

• Chamath Palihapitiya (bullish): Believes Tesla and SpaceX will merge into “Elon Corp” by end of 2025 or mid-2026, creating a single asset that will trade at a valuation premium justified by Elon’s future innovation pipeline.
• Jason Calacanis (bullish): Argued Tesla’s PowerWall with compute inside could be the next distributed data center play; Tesla receives the same “Elon variable” that causes markets to value his companies at 2–4x peers.

Cross-check:

• Price: $428.00 (P/E ~340–376x TTM; mkt cap $1.61T). Next earnings: 2026-07-29.
• Recent headlines: Tesla at $428, mkt cap $1.61T, P/E ~340-376x — extremely elevated multiples consistent with hosts’ framing of Elon premium. No Tesla-SpaceX merger announced as of May 2026.
• Inconsistency: Chamath said the merger would happen “probably by the end of the year” (2025) or “middle of next year” (2026) — as of May 2026 no such merger has been announced.

Topics discussed

Elon Web Services (EWS) — xAI leases Colossus 1 to Anthropic

Summary: xAI leased all of its Colossus 1 data center (H100 GPUs, 300+ MW) to Anthropic, addressing Anthropic’s compute constraints. Hosts discussed this as the emergence of “Elon Web Services” as a fourth hyperscaler alongside AWS, Azure, and GCP. Brad Gerstner estimates the deal generates $4–5B of incremental revenue for xAI/SpaceX in 2026. The deal also subsidizes xAI’s Grok training costs, which had been a major question in the SpaceX valuation story.

Speaker views:

• Chamath Palihapitiya: Elon saw the tea leaves early, built compute at scale, secured power before others, and is now “kingmaking” — this is a valuation reinforcement for SpaceX because it blunts the bear case around delayed orbital data centers.
• Brad Gerstner: The deal solves the biggest question in the xAI/SpaceX valuation story — what if xAI spent ahead of its own revenue — and Gerstner estimates $4–5B of incremental EWS revenue in 2026; he expects more such announcements and calls it a “material part of their story.”
• David Sacks: The deal fixes xAI’s balance sheet problem: xAI had huge losses from training clusters without matching revenue because it lacked a coding product; now it can monetize the capacity while XAI catches up competitively.
• Jason Calacanis: EWS could extend to distributed compute in Tesla PowerWalls and Starlink-connected homes, and the next announcements will move from factories and data centers to consumer hardware.

Potential impact: Hosts discussed this as establishing a fourth hyperscaler and potentially re-rating SpaceX at 40–50x revenue on its IPO road show; the deal was also seen as an accelerant for Anthropic’s revenue trajectory toward $100B ARR by year-end.

Anthropic’s explosive ARR growth and the AI monopoly question

Summary: Anthropic grew ARR from $10B to $30B in Q1 2026, then from $30B to $44B in April, described as unprecedented even by Silicon Valley standards. David Sacks warned this trajectory, if sustained 18 more months, would make Anthropic “the biggest monopoly in human history.” Brad Gerstner pushed back, saying the company is still a fragile startup and competition is robust.

Speaker views:

• David Sacks: Anthropic is on a 10x annual growth rate — tripling in Q1 then accelerating in April — and used the Rockefeller “safe oil” analogy to argue that safety rhetoric is being deployed, consciously or not, to build a regulatory moat around what could become an unprecedented monopoly.
• Brad Gerstner: Called the monopoly framing premature, noting Anthropic on a GAAP basis is doing about the same monthly revenue as OpenAI, Google is also generating substantial AI revenue, and the last thing he wants is DC picking winners at the “starting line of AI.”
• Chamath Palihapitiya: Agreed the trajectory is exponential and a trillion-dollar ARR outcome is logically possible, but noted Dario himself forecast combined AI market leader revenue of ~$1T in 2029 — cautioning they may be “over their skis” on a single-company forecast.

Potential impact: If Anthropic hits $1T ARR it would exceed the combined revenue of the current Mag 7; the regulatory and antitrust implications would be unprecedented, and hosts debated whether proactive regulatory engagement or market competition was the right check.

”FDA for AI” executive order — White House AI regulation debate

Summary: The New York Times reported the Trump White House was considering an executive order to create a review process for new AI models, analogous to an FDA. Kevin Hassett of the NEC confirmed a study was underway. Sacks and Gerstner both disputed the framing, saying they spoke to Hassett and he does not support a pre-approval regime; a statement from White House chief of staff Susie Wiles was cited as largely shooting it down.

Speaker views:

• David Sacks: Called the FDA framing “largely fake news” originating from a commentator rather than a senior official, and said the White House’s actual concern is hardening government systems against cyber-capable AI models — not creating a pre-approval regime that would “squash innovation.”
• Brad Gerstner: Said he spoke to Hassett the evening the clip aired and Hassett confirmed he was only describing a coordination and notification process, not an approval regime; Gerstner supports better government capacity for fast cyber review but opposes any Washington approval process.
• Chamath Palihapitiya: Said some form of AI oversight is coming regardless of administration — the political vibe shift has already happened on Main Street — and the root cause is the tech community’s failure to articulate or deliver broad-based societal benefits.

Potential impact: Hosts discussed that a genuine pre-approval regime would slow frontier model releases, benefit entrenched labs over challengers, and potentially replicate the dynamic that stopped US nuclear reactor construction — Sacks drew an explicit parallel to activist campaigns that blocked fission plants 30 years ago.

AI ROI debate — is the productivity payoff showing up yet?

Summary: Brad Gerstner cited 200bps of S&P 500 operating margin expansion (from ~11.8% in Q1 2024 to ~13% now) and hyper-accelerating cloud revenue as evidence AI ROI is materializing. Chamath countered there is “not a scintilla of evidence” AI has specifically lifted S&P 500 margins, attributing expansion to financial engineering and the post-COVID headcount reset. Sacks sided with Gerstner but acknowledged the question is not fully settled.

Speaker views:

• Brad Gerstner: Pointed to AWS +28%, Azure +39%, GCP +63% revenue growth with only ~3% headcount growth at Mag-5 over three years; also cited enterprise examples (Nike imagery, DoorDash food photos) of AI replacing previously labor-intensive workflows at lower cost and higher ad effectiveness.
• Chamath Palihapitiya: Argued there is no evidence in global GDP, global productivity, or S&P 500 profit margins that AI is specifically causing margin expansion, and said a “reckoning moment” is coming in roughly 500 days when enterprises must show measurable ROI — X spent, Y returned with Y greater than X.
• David Sacks: Noted unemployment has remained near historic lows (~4.2%) even as efficiency gains accrued, and recent data shows college graduates finding it easier to get jobs — suggesting AI is not yet causing the feared displacement while productivity benefits are beginning.

Potential impact: The debate has direct implications for whether current Mag-7 and hyperscaler valuations are justified; Chamath’s “500 days” thesis implies a potential market inflection point in late 2027.

SpaceX IPO valuation thesis

Summary: Brad Gerstner estimated SpaceX will trade at 40–50x revenue on its IPO road show, implying a ~$2T market cap based on $40–50B in projected 2027 revenue. He argued this premium is justified because Elon is the only person on the planet with a pipeline of innovation across launch, Starlink, hyperscaling, orbital data centers, and models.

Speaker views:

• Brad Gerstner: The EWS deal solves the biggest valuation question (what if xAI spends ahead of revenue) and the five-layer SpaceX cake — launch, connectivity, compute hyperscaling, space data centers, and applications — commands a revenue multiple far above Mag-5 peers trading at 25x earnings.
• Chamath Palihapitiya: A Tesla-SpaceX merger into “Elon Corp” is likely by end of 2025 or mid-2026, which will “break everybody’s brains” by creating one unified asset trading at a valuation premium that is logically explainable by Elon’s unique innovation pipeline.

Potential impact: A $2T SpaceX IPO would be the largest in history; hosts implied it would further concentrate wealth and set a new benchmark for revenue-multiple valuations of innovation-driven conglomerates.

Data center protests and activist opposition to AI infrastructure

Summary: Brad Gerstner flagged that nearly 50% of the ~9 gigawatts of data center capacity expected online in 2026 is being protested, drawing a parallel to organized activist campaigns that stopped US nuclear reactor construction 30 years ago. He argued these are not organic community protests but highly coordinated national groups spreading misinformation about water and electricity costs.

Speaker views:

• Brad Gerstner: Argued electricity costs are actually going down in Texas (which is building the most data centers) and going up in New York and California (which aren’t), and called the “data centers raise electricity bills” narrative a boogeyman that needs to be confronted directly.
• Chamath Palihapitiya: Said protests and regulations are partly deserved — the tech community has done a “horrible job” articulating AI’s broad societal benefits and investing in communities, which is creating political antibodies.

Potential impact: If a significant portion of planned data center capacity is delayed or cancelled, compute constraints on frontier AI labs would intensify, potentially slowing revenue growth and extending the period of supply-constrained model deployment.