Everyone Is Still Undersizing the AI Market | Eric Vishria

21 insights saved from this video by @business

@business· How Companies Win
Expert warnings are right 19 times out of 20 or 99 times out of 100. Venture winners live in the exceptions.
@business· How Companies Win
Many SaaS companies quadrupled in size after 2021, yet their valuation multiples collapsed from about 30x to about 6x.
@business· How Companies Win
A company could grow fourfold while its valuation multiple fell sixfold, leaving it worth less despite reaching break-even.
@business· How Companies Win
In 2016, Geoffrey Hinton's technically correct radiology forecast still missed the real-world conditions blocking replacement.
@business· How Companies Win
A working hardware part is only the beginning: physics, dozens of vendors, and roughly 14 bring-up stages remain.
@business· How Companies Win
IPO windows belong to specific company types and market conditions. Miss one, and private employees may lose meaningful liquidity.
@business· How Companies Win
AI teams are building sand castles, not permanent castles. New model capabilities can obsolete their work within months.
@business· How Companies Win
Databases were sticky because migration was painful. AI weakens that lock-in without eliminating the need for databases.
@business· How Companies Win
Frontier model capabilities can change roughly every four weeks, introducing emergent properties that reshape what applications can do.
@business· How Companies Win
AI capabilities have a jagged edge: strengths and failure modes are uneven, unintuitive, and constantly changing.
@business· How Companies Win
The 2014 belief that AWS would consume enterprise software was substantially wrong: Snowflake, Confluent, Elastic, Databricks, and Datadog became huge businesses.
@business· How Companies Win
The AI investment debate is not just models versus applications. It is about where durable moats will actually exist.
@business· How Companies Win
Inference infrastructure looks like commodity resale until optimization expertise transforms performance, throughput, and profitability.
@business· How Companies Win
Tesla and Waymo trained vehicle-controlling models with data collected by fleets of their own autonomous vehicles.
@business· How Companies Win
The next valuable AI companies may become enterprise sherpas, translating rapidly changing capabilities into usable solutions.
@business· How Companies Win
Language models began with writing, editing, and marketing copy before broadening. Robotics may follow the same path.
@business· How Companies Win
AI models now generate code for CPUs, exposing an opening for new architectures without legacy CPU constraints and accumulated baggage.
@business· How Companies Win
Models translate compute into intelligence. If demand for intelligence is unlimited, demand for compute keeps rising.
@business· How Companies Win
The AI competition is not about who can vibe-code. It is about which companies adapt to a changed basis of competition.
@business· How Companies Win
AI is a major new workload constrained by communication, making specialized AI chips a durable category, not a niche experiment.
@business· How Companies Win
Leaders entering AI businesses should invert their effort: prioritize transformation and spend less time operating the legacy business.

More from Invest Like The Best

See everything from Invest Like The Best

Post on X

Keep what you found in Everyone Is Still Undersizing the AI Market | Eric Vishria

A free account saves any of these cards to your Collection, and Korva resurfaces them so you actually remember.