RBC 'Net Add: SPCX, META, UBER

Research evidence excerpt

RBC 'Net Add: SPCX, META, UBER

ile and then

its ability to embed itself in a developer's IDE allowed for efficient lab-agnostic model allocation to

minimize consumption cost. With this new model, we'd expect it to have aspects of Composer (Cursor's

initial frontier model designed for coding) like multi-file editing, but the incorporation with SpaceXAI's

compute should allow for a much larger, ground up model not reliant on any other labs (estimated ~1.5

trillion parameter) that should rival anyone at the true frontier and transitions Cursor's capabilities from

a coding assistant to an actual developer-level capability through RL, trial and error evals etc that are

critical for more deeply embedded argentic loops (ie lower churn risk).

META, GOOGL: Introducing Muse Image: Image Generation Built for Your World (Meta). Our checks

have been finding improving receptivity to META's content creation tools for the past year and so while

this appears a meaningful step forward, we're especially curious about Muse Image's potential to plug

into the recommendation engine for an effective reverse-engineering of conversion for advertisers,

particularly those that are more lacking in resources to create the content permutations that META's

newer models excel at. Consistent with our recent note on a possible future incremental vector to

META's advertising growth, we believe at least one primary portion of MSL's value creation could

be for augmenting, expanding and/or even creating SMB's on META with potential for advertising

growth on the back of META's MSL models offering pre-validation of unexpressed demand which human

intelligence could not detect.

UBER, LYFT: Tesla's Robotaxi push tests new blueprint for scaling fast (The Information). Reports hitting