Supreet

VP of Product @WiserInc, helping retailers and brands stay competitive on pricing.

[Supreet @supreetsharma]

@aleabitoreddit right now my portfolio is

Hyperscaler - 15%
GPU + Customer Silicon - 15%
HBM - 10%
Optical/connectivity/SiPho - 20%
Foundry - 15%
Packaging - 10%
800V - 5%
Neoclouds 5%

what would you size up the most?

[Supreet @supreetsharma]

Great theory, now walk the talk.

Ship the harness. Windows won by running on any x86. Android won by running on any handset. Be the OS for agents, one harness, any model, yours or anyone's.

Abstract the model the way you once abstracted the CPU, and you actually win the layer you say matters.

[Supreet @supreetsharma]

@Gubloinvestor $POWL $TSEM

[Supreet @supreetsharma]

@Gubloinvestor that's the way to go

[Supreet @supreetsharma]

@Gubloinvestor It dropped enough early last week.

[Supreet @supreetsharma]

@Gubloinvestor I think it is just rotation to safer sectors like health care before CPI data drops, SpaceX IPO settles, and, of course, the market adjusts to Japan rate hikes.

So good time to accumulate on positions where there is long-term conviction.

That's take, at least.

[Supreet @supreetsharma]

@DudeWhoInvests Rotation to healthcare before CPI data and upcoming IPO because Semis have ran hot. Great time to accumulate in your strongest conviction longs.

[Supreet @supreetsharma]

@GrindeOptions Rotation to healthcare before CPI data and upcoming IPO because Semis have ran hot. Great time to accumulate in your strongest conviction longs.

[Supreet @supreetsharma]

@Gubloinvestor i was waiting for my DLR transaction to settle 🥲

[Supreet @supreetsharma]

@TheBigBerbowski Fintwit > Bloomberg Chat

[Supreet @supreetsharma]

- How many cars do you see on the road that have FSD?
- How many machines do you see in a factory that are autonomous?
- How many trucks do you see on a highway that have some kind of assistive driving technology?
- How many medical machines do you see just sitting, logging data, not doing anything with it?

Everything will become a smart machine. That is taking some kind of autonomous action on its own without waiting for a human input.

[Supreet @supreetsharma]

@signulll Why would Google be renting compute to Anthropic while also renting from SpaceX

[Supreet @supreetsharma]

@StockMKTNewz $NBIS

[Supreet @supreetsharma]

@RHouseResearch @sk7037 How can this and the fact that SpaceX wants to make data centers in space for inference workloads be true at the same time?

[Supreet @supreetsharma]

@aleabitoreddit @lousyjets I think the TAM expansion and backlog would increase at a higher rate than the rate at which margins can compress.

[Supreet @supreetsharma]

4.6 was best before it got nerfed

[Alex Cohen @anothercohen]

I can't explain why but I fucking hate the way Opus 4.8 writes

[Supreet @supreetsharma]

Both things will be true at the same time for a while.

A lot of user-visible value will move to the orchestration/product layer: workflows, tools, memory, permissions, evals, and vertical context. Users won’t care which model is underneath when the job gets done.

But the model layer won’t become irrelevant. The next leg of capability will be heavily shaped by cheaper inference-time compute — more reasoning tokens per task — enabled by better silicon, and model architectures.

[GREG ISENBERG @gregisenberg]

I didn't cover Claude Opus 4.8 on my pod because I don't think it's MEANINGFULLY better than GPT 5.5 as of May 29th.

We're entering the era where model releases start to feel like iPhone releases. Remember when every new iPhone was a genuine leap? Now it's a slightly better camera and you can't really tell the difference. That's where models are heading. 4.6 to 4.7 to 4.8. Each one is a little different. Nobody can agree if it's better or worse. The benchmarks say one thing, the vibes say another.

The thing that actually matters right now is what's happening around the models. Claude Code shipped dynamic workflows this same week and that genuinely changes what one person can build.

Codex shipped a desktop app with an in app browser that combines coding and knowledge work in one surface. Those are the releases that move the needle for people. The model underneath is becoming interchangeable.

I think we're maybe 6 months from nobody caring which model they're using the way nobody cares which engine is in their Uber. You just want to get where you're going.

When something genuinely changes the game for builders, I'll cover it on @startupideaspod. Opus 4.8 wasn't that. Dynamic workflows was.

I'd rather save you the hour.

[Supreet @supreetsharma]

@edwardlando Repeat after me: all startups are not apps.

[Supreet @supreetsharma]

@garrytan LLM is software

LLM + a harness Maintains, Goals, Memory, Context is also software

All white collar work output is also software

Software is not dead, but its shape has changed drastically.

[Supreet @supreetsharma]

of course, the app layer isn’t dead, but it’s increasingly getting consumed by OpenAI and Anthropic themselves.

They can’t wait for the market to commercialize their models’ intelligence. Most of the massive AI data-center capex depends on them delivering faster ROIC.
As a result, the app layer has turned into a competitive battleground where the model builders are now building directly into it.