Seppmos

Building the Company Brain, one at a time | AI Fluent

[Seppmos @seppmos]

The Shielded Pool

The most important Zcash metric isn't price.

It's the shielded pool where $ZEC becomes cryptographically untraceable.

~1M shielded at Zashi/ZODL launch (late 2024)
~5M shielded today (>30% of supply)
+400% in 18 months.

The people who shield don't sell. 🛡️

[Seppmos @seppmos]

@NousResearch @ghazni1010

[Seppmos @seppmos]

@SecurityElena @naval Yep you're right, they're not identical.

ZEC comes with privacy, BTC does not.

[Seppmos @seppmos]

Zcash: Private Bitcoin

Bitcoin gave us hard money. Zcash gives us private money.

Same 21M cap, same proof-of-work, same halvings, plus the property of privacy that Bitcoin structurally cannot add.

"Bitcoin is insurance against fiat. Zcash is insurance against Bitcoin." @naval

[Seppmos @seppmos]

@heavypulp Insanely well done, cannot stop watching.

[Seppmos @seppmos]

Interesting take by @jswihart on the future of transparent addresses on Zcash.

"Ultimately, I think T-addresses get deprecated at some point and go away. They're not necessary."

[Bankless @Bankless]

INTERVIEW: “ZODL is to zcash:native what Coinbase was to bitcoin:native .”

@jswihart joins Bankless to unpack ZEC's awakening and why @Zcash may be entering its most important chapter yet.

We cover:

- @zodl_app's plan for Zcash
- the rise of shielded ZEC
- privacy in the AI era
- why Bitcoiners are paying attention
- making private money too big to kill

---
TIMESTAMPS

0:00 Intro
4:17 Privacy in the AI Era
6:19 Crypto’s Missing Privacy Layer
9:49 The Political Fight for Privacy
14:42 Can Crypto Retain Privacy?
16:47 From Zashi to ZODL
19:48 The Zcash Reawakening
21:41 Balaji, Funding, and the Mission
24:49 Product Market Fit for Shielded ZEC
27:51 ZODL's Role in the Zcash Stack
29:36 The Zcash Roadmap
32:56 What Is ZEC and is it Undervalued?
36:56 The Shielded Pool Explained
45:12 The $25M ZODL Raise
47:16 Compliance and Shielded ZEC
51:25 Institutions vs Cypherpunk Values
53:58 Personal Privacy in an AI World

[Seppmos @seppmos]

@kantianum @solomon_labs @sptucha YaaS

[Seppmos @seppmos]

Decision markets are for real world impact.

Prediction markets are mostly gambling.

Hence, the TAM of decision markets is at least 10x larger than the TAM of prediction markets.

There are still a lot of unsolved problems to apply decision markets outside of equities and tokens, but just imagine a decision market about the property tax in California...

The trading volume would be insane, and the outcome most likely more beneficial to California as a whole.

[Proph3t @metaproph3t]

https://t.co/xhxJ2x6A1h

[Seppmos @seppmos]

@oxranga Good update.

I never really liked the sUSDv version, glad to hear the staking version will be removed.

Best of luck for the USDv launch.

Love your long term thinking and security-first approach.

[Seppmos @seppmos]

New ATH in Weekly Combined Volume on @AviciMoney

▶️ $1.87M weekly volume
- Up 14% WoW
- Up 98% MoM

$AVICI is ramping up 📶

[Seppmos @seppmos]

@connan_james @MetaDAOProject 👀

[Seppmos @seppmos]

$600K daily volume yesterday alone

[Seppmos @seppmos]

solana:BANKJmvhT8tiJRsBSS1n2HryMBPvT5Ze4HU95DUAmeta is looking good here

- New ATH in weekly volume past week
- On the way to break another ATH this week
- Ownership Coin at $7.7M FDV
- Team is locked in, long term incentive aligned

Peoples Bank

[Seppmos @seppmos]

Love it. Now I wonder, who's to benefit the most out of this expanded economic lifespan of GPUs, other than hyperscalers and neoclouds that can utilize these chips for longer?

And who's about to suffer from this development? Will less chips be sold since existing chips can now live longer, or is token/compute demand still 10x larger than supply, and new chips will be sold the minute they're produced?

[Seppmos @seppmos]

I see, so the TLDR is: if GPUs move over to pre-fill only, there is less friction for them because they are purpose-made for doing pre-fill or just training new models. If GPUs are used to do full inference instead, both combining prefill and decode, they degrade after 3 years. Doing pre-fill only, they could live up to 10+ years?

What about doing training only for those same GPUs? I guess they become economically less valuable after a few years as new GPUs replace them. Hence, training runs become more efficient, and better models can be trained on newer GPUs. At this point, instead of retiring those older GPUs, they are being repurposed and jump in to do pre-fill in the inference side of things. Does this make sense, or am I missing something?

[Seppmos @seppmos]

@grok @tropicalvalue @ezyrider So how is it possible to extend the lifespan of a GPU to 10+ years? If they bread down after 2-3 years at a 70%+ utilization rate?

[Seppmos @seppmos]

I see, makes sense. Now I wonder, what are the general lifespans of GPUs?

Irrespective of their economic value, let's say they're being used for as long as possible until they break or melt, supposing a very high utilization rate of 95%+.

For how many years should such a chip run before it dies?

[Seppmos @seppmos]

Solid explanation. I'd like to double click on expanding a GPUs lifespan. @grok why is it that GPUs suddenly life longer?

- Because they now produce higher economic value than before?
- They're utilized in a different way, running more tailored tasks like pre-fill instead of both (pre-fill and decode) which makes them life longer?

[Seppmos @seppmos]

@HyperForgeAI Pair the more ambitious goal with an ownership coin and this project will thrive + onboard a bunch of loyal users with skin in the game.