AlphaPod — Investment-podcast mention monitor

$NVDA · Why the Markets Are Pricing AI Wrong | Gavin Baker

now playing · $NVDA

$NVDA···bullish· high· posgavin baker

Baker: Nvidia at 10-year low forward PE while compute repricing accelerates hyperscaler cash flows

Nvidia trades at its lowest forward PE in a decade because the market believes it is significantly over-earning, but contracted compute is rolling off at a massive discount to spo…

$NVDA···bullish· high· posgavin baker

Baker: Nvidia at 10-year low forward PE while compute repricing accelerates

Nvidia trades at its lowest forward PE in a decade because the market believes it is significantly over-earning, but contracted GPU capacity is rolling off at a massive discount t…

$ETCHEDbullish· low· posgavin baker

Baker: Etched investor; SRAM accelerator architecture could disaggregate inference and boost ROI on installed compute

Etched's SRAM-based accelerators (Gavin is an investor) could disaggregate inference into prefill, attention, and feed-forward stages, optimizing each on different hardware. This…

$TOGETHERbullish· lowgavin baker

Baker: Together AI among inference clouds growing at frontier-lab pace cash-efficiently

Together AI (referenced as 'together' alongside Fireworks and Modal) is an open-source inference cloud monetizing inference at near-frontier-lab growth rates but burning very litt…

$MODALbullish· lowgavin baker

Baker: Modal among inference clouds growing at frontier-lab pace cash-efficiently

Modal (referenced alongside Fireworks and Together) is an open-source inference cloud growing almost as fast as frontier labs in early days but with minimal cash burn, demonstrati…

$BASETENbullish· lowgavin baker

Baker: Baseten among inference clouds growing at frontier-lab pace cash-efficiently

Baseten (referenced as 'base 10' alongside Fireworks, Modal, Together) is an open-source inference cloud exhibiting near-frontier-lab growth with high cash efficiency, part of a n…

$FIREWORKS-AIbullish· mediumgavin baker

Baker: Fireworks inference cloud growing near frontier-lab pace with minimal cash burn; Nexus product enables AI-native router architecture

Fireworks and similar inference clouds (Together, Modal, Base10) are growing almost as fast as frontier labs in early days but burning very little cash. Their router/RL fine-tunin…

$TOGETHERbullish· lowgavin baker

Baker: Together AI part of inference cloud wave driving open-source token acceleration with capital-efficient models

Together AI is among the inference clouds monetizing open-source inference at scale, contributing to the 'dark matter' of token demand that public markets struggle to measure but…

$MODALbullish· lowgavin baker

Baker: Modal part of capital-efficient inference cloud cohort growing rapidly with low cash burn

Modal is one of the inference clouds (with Fireworks, Together, Base10) growing nearly as fast as frontier labs but with minimal cash burn, representing a new capital-efficient la…

$BASE10bullish· lowgavin baker

Baker: Base10 inference cloud contributing to open-source token demand acceleration

Base10 is among the inference clouds monetizing open-source inference, part of the under-the-radar demand acceleration that public markets miss.

$ANTHROPICbullish· mediumgavin baker

Baker: Anthropic growing strongly and likely FCF positive; third-party data showing deceleration contested by shareholders

Anthropic continues to grow strongly and is almost certainly generating significant free cash flow. The only negative quantitative metric heard is third-party data suggesting its…

$OPENAIbullish· mediumgavin baker

Baker: OpenAI accelerating alongside open source; not slowing despite competitive pressure

OpenAI has accelerated its model releases and growth, countering the narrative that open source competition hurts frontier labs. The combination of OpenAI acceleration, Anthropic…

$ANTHROPICbullish· mediumgavin baker

Baker: Anthropic growing strongly and likely generating significant FCF

Anthropic continues to grow strongly and is almost certainly pumping out significant free cash flow, placing it in pole position among frontier labs. Third-party data suggesting a…

$OPENAIbullish· mediumgavin baker

Baker: OpenAI accelerated back into frontier leadership after compute catch-up

OpenAI has re-accelerated and is back in the game after a period of compute constraint. Game theory dictates no frontier lab will back off compute investment after seeing OpenAI's…

$CRWV···bullish· mediumgavin baker

Baker: CoreWeave one of few non-hyperscalers bringing >500MW/year; contracted compute rolling to spot creates repricing tailwind

CoreWeave has demonstrated ability to energize gigawatts rapidly. Its contracted compute base rolling off into higher spot prices creates a repricing tailwind that accelerates rev…

$META···bullish· mediumgavin baker

Baker: Meta capex intact, Muse 1.1 strong, operating cash flow accelerating

Meta did not cut capex despite market fears; it saw SpaceX monetize compute at massive premiums and signaled similar intent. Muse 1.1 is a strong model release. Operating cash flo…

$SPCX···bullish· mediumgavin baker

Baker: SpaceX compute buildout speed and orbital optionality not priced

SpaceX has brought on compute faster and cheaper than anyone except hyperscalers and CoreWeave. The market ignores the optionality of its 8 GW compute plan (though Baker doubts th…

$MSFT···bullish· mediumgavin baker

Baker: Microsoft brought huge June capacity, operating cash flow accelerating

Microsoft brought on a massive slug of compute capacity in June that didn't even appear in Q2 results. Combined with Meta and Amazon, hyperscaler operating cash flow growth accele…

$SAFE-SUPERINTELLIGENCEneutral· lowgavin baker

Baker: SSI targeting continual learning breakthrough with August model release

SSI (Safe Superintelligence) plans to release a model in August focused on continual/sample-efficient learning. If solved, this could reduce training compute demand but Baker beli…

$COGNITIONbullish· mediumgavin baker

Baker: Cognition index shows AI spend correlates with revenue growth

Cognition's data shows companies spending most on AI (tokens as % of comp) grow meaningfully faster. AI natives spend 20-50% of total compensation on tokens instead of hiring huma…

$MU···neutral· mediumgavin baker

Baker: Memory LTAs create game-theoretic lock-in; breaking agreements risks allocation loss in next upcycle

Memory suppliers (Micron, SK Hynix, Samsung) are shifting to long-term supply agreements (LTAs) with floor/ceiling pricing. Hyperscalers cannot break LTAs without risking allocati…

$MU···bullish· highgavin baker

Baker: Memory LTAs create game-theory lock-in; Micron could rival labs in importance

Hyperscalers are signing long-term supply agreements (LTAs) with memory vendors, prepaying for capacity with floor/ceiling pricing. Breaking an LTA risks permanent allocation loss…

$HYNIX···neutral· mediumgavin baker

Baker: SK Hynix (Highex) same LTA game theory as Micron; hyperscalers locked in

SK Hynix benefits from the same LTA dynamic as Micron. Hyperscalers breaking LTAs would risk future HBM allocations, making memory revenue more durable than market assumes.

$SAFE-SUPERINTELLIGENCEneutral· lowgavin baker

Baker: SSI new lab focused on continual/sample-efficient learning; potential discontinuity in training compute demand if solved

SSI and a generation of new labs are targeting continual learning and sample-efficient training. If solved, models could train on 10T tokens then learn in the wild, potentially re…

$BLACKSTONEneutral· lowgavin baker

Baker: Blackstone and Apollo proposing credit-wrapper structures to memory companies to capture revenue share

Private credit firms (Blackstone, Apollo) are likely pitching memory vendors to put up cash against LTAs in exchange for revenue royalties, mirroring Nvidia's credit-wrapper model…

$APOLLOneutral· lowgavin baker

Baker: Apollo alongside Blackstone structuring memory credit wrappers with revenue participation

Apollo is participating in the same private credit innovation for memory suppliers, offering upfront cash against LTAs for ongoing revenue share, extending the Nvidia credit-wrapp…

$ASML···bearish· mediumgavin baker

Baker: China DUV breakthrough overreacted; 5+ years before hitting ASML orders; EUV gap remains 25 years

China's DUV machine is a phase transition (prop plane to jet turbine) but 25 years behind EUV. Market overreacted; any impact on ASML orders is 5+ years out. Learning-by-doing cyc…

$FIREWORKS-AIbullish· mediumgavin baker

Baker: Fireworks Nexus enables AI natives to customize open models defensibly

Fireworks' Nexus product ingests proprietary data, RL-tunes open models, and routes queries via a three-line integration. This lets AI natives shift 30-60% of token consumption to…

$HARVEYbullish· lowgavin baker

Baker: Harvey legal AI leaned into open-source router architecture pre-acquisition; model for AI-native defensibility

Harvey (pre-acquisition) and Cursor leaned heavily into customizing open-source models via inference clouds like Fireworks, achieving better performance at half the cost of pure f…

$RAMPbullish· lowgavin baker

Baker: Ramp data shows AI-native companies spending 20-25% of comp on tokens; growing faster than peers

Ramp's spend data reveals AI-native companies allocating 20-25% (up to 50%) of total compensation budgets to token inference, correlating with faster growth. This labor-substituti…

$COGNITIONbullish· mediumgavin baker

Baker: Cognition index shows AI-heavy companies growing meaningfully faster; Scott Wu a 'dark horse' Game of Thrones player

Cognition's data shows companies spending most on AI grow faster. The 'Cognition Index' (though not industry-controlled) demonstrates the productivity gains from agentic AI. Scott…

$STRIPEbullish· lowgavin baker

Baker: Stripe charts confirm AI spend correlates with faster growth; supports productivity thesis

Stripe's internal data (alongside Cognition and Ramp) shows companies with highest AI token spend grow meaningfully faster, reinforcing the view that AI drives productivity not ju…

$SPCX···bullish· mediumgavin baker

Baker: SpaceX compute buildout speed and cost advantage not priced; 8GW target likely aggressive but any material progress is upside

SpaceX has brought on compute faster and cheaper than anyone except hyperscalers and CoreWeave/Crusoe. The market applies a large short case assuming spot compute prices collapse,…

$CURSORbullish· mediumgavin baker

Baker: Cursor acquired by SpaceX; 30% token spend of total comp shows AI-native efficiency; Grok Build integration accelerates

Cursor's acquisition by SpaceX combines a leading AI coding agent (30% of compensation spend on tokens, highest heard) with Grok 4.5 and SpaceX's compute advantage. The router arc…

$CRUSOEbullish· lowgavin baker

Baker: Crusoe among elite few bringing >500MW/year; validates non-hyperscaler compute buildout capability

Crusoe's ability to bring on >500MW/year alongside CoreWeave and SpaceX proves that non-hyperscaler neoclouds can execute at hyperscaler speed, supporting the thesis that compute…

$CRWV···bullish· mediumgavin baker

Baker: CoreWeave among elite few scaling GW-class compute annually

CoreWeave is one of only four entities (with hyperscalers, Crusoe, SpaceX) to bring online >500 MW/year, doing so at the fastest pace and lowest cost. Its contracted base rolls of…

$STARCLOUDbullish· lowgavin baker

Baker: Benchmark-funded orbital compute startup validates SpaceX orbital thesis; uses Starlink laser tech

Benchmark's investment in StarCloud (orbital compute, partnering with SpaceX for Starlink laser links) is a sanity check that orbital compute is credible. If Benchmark, SpaceX eng…

$BENCHMARKbullish· lowgavin baker

Baker: Benchmark funding StarCloud validates orbital compute; partners not from Elon ecosystem

Benchmark's investment in StarCloud (orbital compute) at a decent valuation without SpaceX's internal launch cost advantage signals that sophisticated VCs see orbital compute as v…