The AFF BS Meter is running high once again.
@bwilcove launched Internet products during the fiber glut. Everyone using 1999 to wave away the AI buildout is getting the one detail that mattered exactly backwards.
Glass keeps. Silicon doesn’t.
Read his thoughts below:
https://t.co/yvJQxL532p
THE BS METER: 🔴🔴🔴🔴⚪ — Mostly BS
Clean 🔴🔴🔴🔴🔴 for the “connecting the unconnected billions” version.
Satellite can succeed in remote, mobile, defense, maritime, and backup connectivity. It cannot eat fiber or dense terrestrial networks.
The physics breaks first. The economics break second.
Full piece:
https://t.co/2BfqKarkqU
Sovereign AI is mostly theater.
Flags, press releases, and server racks.
BS Meter: Pegged.
The real leverage is in the data.
Full breakdown below ↓
https://t.co/F7rGq6FyRp
We’ve been saying that.
Hands have always been the problem. Humanoid demos are entertaining, but real-world applications live or die on dexterity.
Can it wire a harness? Assemble a watch? Handle soft and irregular objects without breaking them?
Glad the attention is finally moving here. Solutions around dexterity, actuation, tactile feedback, force control, and real-world generalization will be the real unlock.
The asset light model was never about eliminating cost. It concentrated cost in distribution. Whoever acquired customers fastest and cheapest won. That rewarded whoever raised the most capital and blitzscaled to a defensible user base. The moat was never the product. It was the installed base you bought.
AI breaks that logic. If anyone can spin up a comparable product in a weekend, the $200M you spent acquiring users can churn off a single good launch post. Distribution was the moat. Now it isn’t.
Machines need to think like humans… BULLSHIT
The AFF BS Meter: 9/10
We have 8.3 billion humans already. Why build AI to act like more of them?
That’s cloning, not progress.
Real tech doesn’t imitate. It replaces.
Calculators. GPS.
Human-like is a dead end.
Capability is the frontier.
Read the full breakdown below 👇
https://t.co/bEMtEiPATo
Hot take: There hasn’t been a real battery “breakthrough.”
Not because the science is fake, but because none of it survives manufacturing.
The AFF BS Meter: 8/10
The real breakthroughs happen in factories, not labs.
See our full take below ↓
https://t.co/KBrRYEHU5G
@pmitu This is nothing new. This game has been going on forever. Seed was the new A, now it’s pre-seed. It’s just a way for VCs to try to differentiate and raise capital.
You didn’t automate the task. You delegated the decision.
You’re Not Using AI. You’re Delegating to It.
AFF BS Meter: 7/10
The shift isn’t better tools.
It’s giving up control.
And we’re way earlier on trust than people think.
Read the full breakdown below 👇
https://t.co/b32BU9vTCl
We got another one for you.
America Is Sleepwalking Into an AI Disaster
AFF BS Meter: 7/10
Billions of dollars later and we’re still using AI to make broken factories slightly less bad.
Optimization ≠ transformation.
Read the full breakdown below 👇
https://t.co/8Qep3ubalT
@ry_paddy Capital intensity isn’t lower in software, it’s redirected to GTM. The real difference is time and feedback loops. Software lets capital compound quickly. Deep tech doesn’t.
Capital is a commodity. The check clears the same whether it's from a great investor or a mediocre one.
The edge comes from everything else, think networking, pattern recognition, credentialing, and operator judgment. The best VCs have seen 50 versions of your failure mode before you have. That's worth something.
Not that weird once you separate narrative vs. earnings mechanics. SV is underwriting exponential token growth. Public markets are underwriting unit economics + durability.
Right now:
• Tokens ↑ fast, but price/token ↓ faster
• Inference is shifting from scarcity → competition/commoditization risk
• Margins unclear (who captures it: model, infra, or app?)
• Heavy capex + depreciation → near-term earnings pressure
• Hyperscalers may internalize → compress standalone winners
So you get: Private = demand story. Public = who actually makes money from it (still unclear)