CHAPTERS
- 0:00 – 0:30
Meta’s Q2 beat: ads growth masks AI spending fears
Kara breaks down Meta’s strong quarter, driven by a rebound in advertising revenue and profitability. She frames the central tension: investors like the core business strength, but worry about escalating AI capex and unclear near-term payoffs.
- •Meta beats estimates on revenue and profit; shares jump after hours
- •Advertising revenue up 22% year-over-year
- •AI capex guidance rises to at least $37B for 2024
- •Market more tolerant of spending this quarter than last
- •Open question: how long investors will wait for AI ROI
- 0:30 – 1:10
Microsoft’s mixed quarter: cloud growth slows and the stock wobbles
Kara quickly summarizes Microsoft’s results as solid overall but with a key soft spot: cloud growth slightly under expectations. The market reaction reflects investor sensitivity to any signs that AI-era cloud demand isn’t accelerating fast enough.
- •Overall sales and profit growth beat expectations
- •Cloud revenue up 27% but below prior quarter and analyst expectations
- •Shares fall ~3% post-earnings, then partially rebound
- •Sets up a compare/contrast with Meta’s investor reception
- 1:10 – 1:40
Scott’s Meta scorecard: massive scale and accelerating monetization
Scott emphasizes just how strong Meta’s underlying machine is, highlighting sharp EPS growth and improving ad dynamics. He argues the fundamentals (users, pricing, impressions) make Meta uniquely able to fund the AI buildout.
- •Meta EPS up 73% year-over-year
- •Ad impressions up 10% and average price per ad up 10%
- •~3.25B daily active users; ~40% of the planet engaged daily
- •Meta’s core ad business provides a cushion for AI investment
- 1:40 – 2:10
The compute shock: LLaMA 4 training and the capex spiral
Scott explains what’s spooking investors: the implied explosion in compute requirements for next-gen models. He connects that warning to broader fears that AI infrastructure spending could outpace monetization for years.
- •Meta signals AI capex growth will be significant
- •Training LLaMA 4 may require ~10x the compute used for LLaMA 3
- •Investor anxiety centers on how quickly spend scales versus returns
- •AI competitiveness increasingly defined by access to compute
- 2:10 – 2:40
Software vs. hardware: why the market rewards chip suppliers first
Scott outlines a classic cycle where investors rotate to the “picks and shovels” layer when platform economics feel uncertain. He contrasts rising capex and slower revenue growth for big software platforms with explosive revenue growth for hardware leaders.
- •Big software platforms’ capex up ~31% while revenue up ~19%
- •Hardware layer revenue up ~146% with flat/down capex (per his framing)
- •NVIDIA/AMD/Arm seen as safer near-term beneficiaries
- •Post-earnings: Microsoft/Google down while AMD pops after results
- 2:40 – 3:16
Microsoft’s capex jolt and the $200B AI spending spree
Scott highlights the specific Microsoft datapoint that rattled markets: a steep year-over-year jump in capex. He then zooms out to show how unprecedented the collective spending by the biggest players has become.
- •Microsoft capex cited as up ~78% YoY to ~$19B
- •Amazon/Meta/Microsoft/Google combined capex ~ $200B this year
- •Compares scale to U.S. homeland security spending to underline magnitude
- •Investor focus shifts from growth to capital intensity and payback timelines
- 3:16 – 3:48
Will anyone make money? The ROI anxiety behind the AI arms race
Kara questions whether the industry can recoup these massive investments, noting that clear profit centers are still limited. She argues companies may have no choice—unlike the metaverse push, AI is a fundamental computing shift.
- •Central concern: uncertain payoffs despite soaring spend
- •Kara suggests OpenAI is among the few clearly making money right now
- •A VC anecdote underscores widespread monetization skepticism
- •Kara’s view: companies can’t opt out; AI is a critical platform transition
- 3:48 – 4:19
Why everyone is ‘all in’: captive distribution and cheap capital advantages
Scott argues that today’s AI revenue may be small versus spend, but it’s still ahead of early web-era monetization—and incumbents have huge built-in distribution. He also frames capex as strategic: firms with cheap capital spend aggressively to widen moats.
- •AI resembles early web hype, but monetization is stronger than the 1990s web
- •Incumbents can deploy AI to existing captive audiences
- •Cheap capital/strong market caps enable aggressive capex as a competitive weapon
- •“Turn on the spending jets” to pull away from rivals
- 4:19 – 4:58
AI everywhere: earnings-call buzz as a signal of a tidal wave
Scott points to how broadly AI has permeated corporate strategy, extending far beyond the tech sector. The statistic about AI mentions on earnings calls is used as evidence that AI is reshaping business priorities across industries.
- •~40% of S&P 500 companies discussed AI on earnings calls
- •Only ~20% of S&P 500 are tech, implying broad adoption pressure
- •AI mentions up from ~1% five years ago
- •AI framed as a ‘tidal wave’ influencing all sectors
- 4:58 – 5:29
Who should build vs. rent AI? The emerging guidance for non-leaders
Scott predicts a bifurcation: dominant players will earn strong returns, while many others should avoid heavy capex and instead buy AI capability. He cautions against second-tier firms trying to build proprietary LLMs without scale advantages.
- •AI spend is ‘well spent’ for some—mainly the biggest incumbents
- •Second-tier/non-tech firms should ‘rent’ AI rather than build infrastructure
- •Warning against companies attempting to develop their own LLMs without scale
- •Capex discipline becomes a differentiator outside the AI leadership cohort
- 5:29 – 6:08
Valuation reality check: $20B AI revenue vs. $3T market-cap lift
Scott quantifies the disconnect between current AI application revenues and the market value created around AI expectations. He argues implied multiples look unsustainable unless revenues compound extraordinarily fast for years.
- •Estimated current AI app revenue ~ $20B
- •AI has contributed to ~ $3T increase in market capitalization (per his estimate)
- •Implied ~150x revenue multiple for the industry notionally
- •To justify pricing, revenues would need to grow extremely rapidly for a long period
- 6:08 – 6:19
Closing consensus: it’s an unavoidable AI arms race
Kara and Scott converge on the idea that, regardless of valuation risk, companies feel compelled to keep spending to stay competitive. The segment ends with the blunt framing that opting out isn’t viable.
- •Kara: hard to see how companies avoid these spends
- •Scott: AI is an arms race; they ‘have to do it’
- •Shared view: competitive pressure forces continued capex escalation
