Mag 7 sell-off, Wiz rejects Google, UBI, Kamala in, China's nuclear buildout, Sacks responds to PG
2024-07-26 spoken.md · speaker-labeled ▶ watch ← E188 all episodes E190 →
Every number here is replayed from score_events — the same ledger the pool ranks on. Decay is what the 126 ideas nobody mentioned gave up this week; it applies only when an episode is processed.
Tier crossings
conviction thresholds crossed by this episode — 65 / 45 / 15 · ideas born here show where they landed. why 65 isn't always green
Who moved the board
each voice's force on conviction this episode — supports and opposes, weighted exactly as the replay applied them · share = % of this episode's movement
What got argued (18 ideas)
ordered by how hard each idea moved · quotes are verbatim from the transcript, timestamps deep-link into the episode
GCP just printed $10.3B of quarterly revenue against $8B a year earlier and $1.2B of operating profit, and at a normalised 20-30% cloud margin becomes a $20-30B/yr free-cash-flow engine on its own — against Alphabet's current ~$60B total. Its first seven years track AWS's almost exactly. With the core consumer search business carrying the risk, cloud plus AI tooling is where Alphabet's growth actually comes from, and the market is underrating GCP as a second money printer alongside YouTube.
in this past quarter, they're running at a $1.2 billion operating profit out of GCP. So, if you kind of think about what cloud margin should be over time and kind of call it 20 to 30% margin, this cloud business could be generating 20 to $30 billion a year of free cashflow.
Fox News's profitable niche depends on being the one network that serves conservatives, but Rupert is a neocon and neoconservatism is losing the Republican party to Trump/Vance populism. Firing Tucker, its highest-rated host, and campaigning against Vance on the ticket are symptoms of a market-position problem: the network is drifting away from the audience it monetises. Bearish the Fox/News Corp franchises regardless of how the Murdoch succession case resolves, and a liberal-sibling outcome would be worse still.
Bricking 8.5 million Windows machines is not a technical blip but a trust event for a company whose whole product is trust: enterprises that did not even know they were running the agent are now auditing it out, and high-profile customers (Musk's companies, Craft's own IT) are removing it. Bearish CrowdStrike's renewal and net-new pipeline versus rival endpoint and cloud-security vendors.
Sacks discloses he owns Palantir and calls it a good investment on the back of the very thing critics use against it — that it sells deep into the national-security state. Bullish PLTR as a sole-source government software franchise.
The multiples that we have right now are not that insane. I mean, I think the long-term median software multiple is around a seven. So, you know, you see here the mid-growth median 7.7. We're about tracking at the historical pre-COVID mean, and we had that really frothy bubbly period in 2020, and 2021, and that's clearly over.
Episode digest
written during extraction and stored in data/extractions/ep189.json — the auditable source of truth, including everything market-adjacent that did not earn a capture
Dense market episode. Wiz rejecting Google's $23B: Chamath frames cloud security as structurally under-provisioned and expects Amazon and Microsoft to want the same asset, Friedberg puts a $100B Palo Alto-style outcome in play within a couple of years and predicts Google keeps hunting sizeable public M&A that can clear antitrust — both Sacks and Friedberg name deal-blocking risk as a reason Wiz walked, reinforcing the antitrust-chill thesis. Friedberg lays out GCP's $10.3B quarter and $1.2B operating profit as a future $20-30B/yr FCF engine tracking AWS's curve, while Sacks says software multiples are back at the pre-COVID mean and cloud growth is still strong. On the Mag-7 sell-off the panel converges on rotation, not regime change: Chamath calls the market priced to perfection with a bad consumer-credit setup into the fall, Friedberg calls it a mini-AI-bubble deflation and a rebalancing into rate-cut trades, Sacks brushes off a single red day. Sacks discloses he pulled CrowdStrike from Craft's IT and does not trust the company, and separately discloses a long PLTR position; he also argues Fox News's neocon drift is misaligning it with its populist audience and that this is bad for business. Kamala's 48-hour coronation is treated as the only path all three besties expected once Biden went, with Sacks effectively reversing his 'Biden stays the nominee' call. Friedberg's science corner on China's pebble-bed Gen-4 reactor is the strongest energy capture of the episode — China at 3TW heading to 8.7TW by 2050 versus the US doubling to 2TW, at a third the cost per kWh — but its explicit framing is 2050/2060, so the decades-out industrial-competitiveness thesis stays in the digest rather than becoming an idea; his in-window claims attach to the existing China-cheap-power, AI-power-demand and US-nuclear-regulation ideas. Sam Altman's UBI study (null results on health, work and education) got a long, unanimous thumbs-down but produced no tradeable claim beyond Friedberg's hypothetical inflation argument.