# DeFi — X 热门讨论 (2026-09-28 08:43 UTC)
## @base (Base) · 09-28 08:17 · ♥126 ↻25 💬37 Last Week on Base: DeFi all-time high, Tokenized Stocks on Aave, and more https://x.com/base/status/2104485619585949759
## @NicolaBlackwo12 (Nicola Web3) · 09-28 04:05 · ♥79 ↻2 💬84 Running agent jobs across chains only works if the trust layer travels with them
With @termix_ai, AACP targets Base for low cost verification, BNB Chain for broader DeFi activity, and Ethereum for higher value settlement
The architecture separates identity, reputation, escrow, staking, disputes, fees, proof checks, and TEE verification into dedicated contracts
That means each part of the job lifecycle can be enforced onchain instead of relying on one oversized contract
Would you rather see agent infrastructure expand chain by chain or focus deeply on one network first? https://x.com/NicolaBlackwo12/status/2104422094742356297
## @ayoadhe (Abdulbasit) · 09-28 04:22 · ♥78 ↻8 💬66 bitcoin was built to be money without a middleman.
starknet was built to make onchain applications more scalable and capable.
for years, those two worlds have largely operated in their own lanes.
that is what makes strkBTC interesting.
strkBTC is a bitcoin-backed token on Starknet that represents BTC on Starknet, backed 1:1 by bitcoin.
in simple terms:
you have BTC on Bitcoin.
you bridge it into Starknet through a supported route.
you receive strkBTC.
your BTC is still the asset backing what you now hold, but strkBTC gives that value a way to interact with the Starknet ecosystem.
bitcoin is incredibly useful as an asset to hold.
but holding isn’t the only thing people want to do with capital.
sometimes you want to trade.
sometimes you want to put your assets to work in DeFi.
sometimes you want to access crypto lending or borrowing.
sometimes you simply want your bitcoin to be usable across applications instead of sitting in one place.
strkBTC is designed to give BTC that flexibility on Starknet.
then there is the privacy question.
public blockchains give us transparency, but transparency doesn’t always mean you want every part of your financial activity exposed in the same way.
strkBTC gives users a choice.
you can use it publicly, or you can choose to shield it when you want more control over what is publicly visible. shielding is opt-in, not automatic.
so when you look at strkBTC, don’t just think:
“another BTC token.”
think about the problem it is trying to solve.
how do you take one of the most important assets in crypto and give it a more useful place inside an application ecosystem, while also giving users more control over their onchain privacy?
that is the bigger idea.
BTC brings the value.
Starknet brings the environment.
strkBTC connects the two.
and suddenly, your bitcoin isn’t only something you hold.
it becomes something you can actually use.
@Starknet
as always, using bridges and DeFi comes with smart contract, bridge, liquidity and market risks. do your own research and only use products you understand. https://x.com/ayoadhe/status/2104426435020550474
## @wnyckyoung (Web3 Young) · 09-28 05:45 · ♥87 ↻12 💬13 Bullish conviction matters to DeFi when it becomes deeper liquidity and healthier borrowing demand. If collateral discipline loosens instead, that is leverage masquerading as strength. https://t.co/V0aJ2NLjPZ > 引用 @saylor: Ciao, ₿ears. https://t.co/zXsgpCztu9 https://x.com/wnyckyoung/status/2104447506880893238
## @DamiDefi (Dami-Defi) · 09-28 08:01 · ♥91 ↻12 💬7 $TAO COULD BE UNDERVALUED IF AI INFERENCE COSTS BECOME THE NEXT BIG TRADE.
SOMA, Bittensor Subnet 114, just demoed a prompt compressor cutting token counts by 76%.
→ ~15% average coding cost savings → GitHub Copilot + Codex early access → Already being used in open-source workflows
Less context processed = lower inference costs.
If Bittensor keeps solving expensive AI bottlenecks like this, $TAO's utility thesis gets stronger with every subnet that ships. > 引用 @opentensor: SOMA cut a prompt’s token count by 76% live, reducing the text you pay AI models to process.
On the latest Novelty Search, Subnet 114 presents its compression tool for coding agents, already in early access for Copilot and Codex.
The team reports ~15% lower inference costs on average across coding tasks. @say_gm_ has integrated SOMA’s open-source compressor, and Dendrite uses it in daily development.
Hosted by @const_reborn
Full episode in the first comment https://x.com/DamiDefi/status/2104481684699930688
## @FabiusDefi (Fabius DeFi) · 09-28 05:52 · ♥77 ↻5 💬27 There’s a shift in DeFi that I’ve been talking about lately:
Curators are gradually becoming a layer of their own.
You can already see it clearly on Morpho, where names like Gauntlet, Steakhouse, Galaxy, Wintermute,... are increasingly acting as strategy/risk managers between protocols and depositors.
But most vaults still mainly solve one problem: capital allocation.
@ProsperTicker (built on @pharos_network) caught my attention because it’s trying to add another layer on top: strategy discovery + market conviction.
With already 20+ Vaults live, ranging from trend-following to structured allocation, equities, gold and AI infra.
💡Btw, there's also a p{VAULT} Trading Challenge with 100K $PROS prize pool just went live a few days ago (worth checking out imo).
The flow itself is still pretty familiar:
Curator → Strategy → Vault
If you want direct exposure to the strategy > deposit into the Vault and receive Vault Shares, which are NAV-linked and redeemable against your proportional share of the Vault assets.
But alongside that same strategy, there is also a separate p{VAULT} market.
p{VAULT} has its own market price, driven by supply and demand, and it can move independently from the Vault itself.
It is not a Vault Share, does not represent ownership of the Vault assets, and does not track the strategy’s NAV or profits.
What I find interesting is that Prosper lets you observe 2 things at the same time: actual Vault performance on one side, and market conviction on the other.
A Curator could be executing well while the market still pays very little attention.
Or the opposite could happen, where a p{VAULT} attracts attention much faster than what the strategy itself has actually proven so far.
That creates a dynamic traditional vaults rarely turn into a separate market.
For me, the interesting part of Prosper is not simply just another vault platform.
But they’re testing whether verifiable strategy performance can become a new market primitive of its own 💎 https://x.com/FabiusDefi/status/2104449224612131144