ZK Zhao https://paragraph.com/@zkzhao undefined Tue, 27 Oct 2026 06:39:59 GMT https://validator.w3.org/feed/docs/rss2.html https://github.com/jpmonette/feed en ZK Zhao https://storage.googleapis.com/papyrus_images/2f0964b0d6c5de3f0004d33e6a507975 https://paragraph.com/@zkzhao All rights reserved <![CDATA[CryptoxAI Landscape, 2024.02]]> https://paragraph.com/@zkzhao/cryptoxai-landscape,-202402 geMsXe8A9SbHRKU4SDiq Sun, 11 Feb 2024 01:36:36 GMT

1. Types

So many cryptoxAI projects, how to understand them? (e.g., Bittensor, MyShell, Ritual, Bittensor, Modulus... what are their diff?)

My take:

  • Gen-AI related: end goal is to create things like chatgpt, midjourney...

  • On-chain ML related: end goal is to put ML models on chain (esp on EVM-chains), so app are more related to optimization, credit-scoring etc.

  • Dev component: things like data and hardware/GPU that are key supporting elements to AI developement (but not related to the client-side)

2. Projects

Specific project links to-be-add 🤣

2.1 Gen-AI

  • App: things like chatgpt where applications are the focus

  • Middleware: things like GPTs, right now most are things related to agent-level

  • Chain

2.2 On-chain ML

  • App: mostly things related to on-chain Defi, using ML to optimize yield, risk mgmt etc.

  • Middleware: how to serve ML onchain; most are ZKML-related

  • Chain

2.3 Dev Component

  • Data: how to acquire data to train, fine-tune AI models; data marketplace, monetization, labelling etc.

  • Hardware: how to acquire GPUs to train, serve AI models (mostly DePIN-related)

3. Thoughts

(1) Why CryptoxAI

recom checking about; tldr:

  • some issue are very suitable for crypto experiment; for example, GPU shortage

  • crypto is a good way to support open dev platform, can be a good alternative to centralized, close-source dev platform like OpenAI (think about Twitter/FB platform back in the old days)*

*my concern is that, AI dev platform is a bit different from social platform; that the network effect is different, it's sit between AWS and Twitter/FB. Also how data will be shared / used by these commercial platform is TBD

(2) Interesting Projects

Bittensor, Shinkai, FedML, Bagel

(3) Bearish view

still, I have some bearish view on some sector / project, can't share publically, but already share to some of my vc and builder frens

4. Resources

What are some good resources to learn CryptoxAI?

(1) Landscape overview

(2) Research / Discussion

Finally, thanks ash, casey, chao, jacob, laobai, juylan for feedback and discussion

]]> zkzhao@newsletter.paragraph.com (zkzhao.eth) <![CDATA[Orderbook DEXs: the next exit for crypto?]]> https://paragraph.com/@zkzhao/orderbook-dexs-the-next-exit-for-crypto 4RGB4VisPGKwzooarBh1 Thu, 22 Jun 2023 15:05:34 GMT
Image

tldr

  • 3 major types for orderbook dex: rollup-based, fully-onchain, high-perf

  • dydx v4, hyperliquid, vertex, aevo, orderly, apex are doing some interesting new developments in this space

  • security, convenience to existing exchanges and GTM strategies are the things that matter most for orderbook dex exchanges.

1/ Landscape

There are many orderbook exchanges. Understanding them involves grasping a performance-decentralization trade-off (the diagonal); increased performance implies more centralized design. Here's breakdown:

- CEX: Max performance, highly centralized (@FTX_Official)
- fully on-chain: Max decentralization, performance-limited; unfit for orderbook due to high gas fees (@GMX_IO)
- roll-up based: Off-chain orderbook for performance; settlements handled on-chain in batches for security & lower gas fee (@dYdX v3, @zkLinkorg @OrderlyNetwork)
- high-perf chain: A tailored chain to cater to the high-performance demand of the orderbook (@dydx v4, @Injective_, @vegaprotocol, @SeiNetwork)

As for metric comparison, I recommend @DefiLlama: https://defillama.com/protocols/Derivatives

2/ Interesting developments you might have missed

2.1 @dydx

v4 launch coming soon (Q3?). I admire their daring vision of creating a fully decentralized, performant exchange, which is fully open-sourced.

2.2 @HyperliquidX

A high-performance chain that seems to be developing faster than dydx 😂 (mainnet now)

- all actions are on-chain, for example, https://app.hyperliquid.xyz/explorer/address/0x63c621a33714ec48660e32f2374895c8026a3a00
- no gas for orders, curious about their DDoS defense 🤔

2.3 @vertex_protocol

one of the most feature-rich rollup-based orderbook DEX
- combines orderbook & AMM (Better composability?)
- on-chain risk management (no more liquidation by hayes 😂)
- unified margin across spot, perps & money markets

2.4 @aevoxyz

options + perp + otc; a first in the market; provides portfolio margin with strong backing from @ribbonfinance, @paradigm. They've executed very well since launch

2.5 @OrderlyNetwork

offers trading-infrastructure-as-a-service, a different go-to-market strategy. Building an orderbook DEX is challenging; if Orderly simplifies it, we may see some amazing growth similar to GMX, but strong security is also needed.

2.6 @OfficialApeXdex

while dydx has moved away from starkex, ApeX is still building on it w/ good improvs
- cross-chain deposit (eth, bsc, arb, polygon, op, avax)
- smart lp
- app
- referral

might be the most user-friendly orderbook dex atm.

2.7 @StorkOracle

A new oracle infrastructure. There're many oracles, but few see meaningful adoption outside their token 😂. And yet, Stork is adopted by @vertex_protocol and @OfficialApeXdex, as orderbook requires a faster oracle; love to see how they'll grow in future.

3/ What matters for adoption?

Due to my personal exp, I get asked a lot on what features matters for exchange adoption 🤣. exchange has innovative designs, but few gain traction. Here are my thoughts:

3.1 Security

How secure can a DEX be? My fav example is @zkLinkorg; it shows even when it is down, funds are still safu. While many DEX claim they're safe, but in reality not, campaigns like this are proof-of-security and make good players stand-out.

3.2 Low Cost & Convenience

Can users deposit from multiple chains? What's the order placement cost? How about fees, rewards, and rebates? Since users mostly come from existing exchanges, orderbook DEXs must offer a similar UX to encourage migration.

3.3 Go-To-Market Strategies

While product features are often discussed, they alone don't ensure success. Why is @dydx v3 the biggest among multiple exchanges built on StarkEx, despite having similar features?

To some extent, exchange is a network product (like twitter, fb) and liquidity is a cold-start problem. How to bootstrap liquidity? Choose perps vs. spot? How to stand out from competitors? @Bybit_Official and @bitgetglobal are good case studies if you're familiar with their journeys.

I'll leave a chart w/o comment 😉

Image

4/ Future

We've seen remarkable progress in this space compared to the last cycle. I believe the journey doesn't stop here. Hats off to the teams shipping during the bear market, hoping we can finally free ourselves from figures like SBF and Gary 😅.

And that's it, thanks ... for feedback

and feel free give any advice, or point out any error or things I missed.

]]> zkzhao@newsletter.paragraph.com (zkzhao.eth) <![CDATA[Finding the Balance: Fair Terms for Raising and Building Crypto Projects]]> https://paragraph.com/@zkzhao/finding-the-balance-fair-terms-for-raising-and-building-crypto-projects Av9vvfxoL5CBD1A8oot4 Sat, 22 Apr 2023 11:50:34 GMT Lately, I've had a thought bouncing around in my head, and it's about time I let it out. What is the fairest way to raise and build a project in the world of Web3 and cryptocurrencies? i.e., “Fair Raise”. Brace yourself because this is a controversial topic. Let's dive in!

1. Issues and Why It Matters

  • In the fast-paced world of crypto, it's essential to establish good terms for both sides to raise money. These "fair terms" are often up for debate, and there are a few key points to consider:

    • a) Greedy Terms: This approach involves raising a lot of money and unlocking tokens aggressively. While I have personally benefited from such terms, it may not be the fairest way to incentivize a project's long-term success.

    • b) Fair Launch: On the other end of the spectrum, there's the fair launch, where the team behind the project receives no tokens. But without financial incentives, will they remain committed to the project's success?

    • c) Communication Breakdown: It's frustrating when a project's team is unresponsive and later informs us that they've failed and used up all the funds. This situation leaves investors feeling hurt and betrayed.

  • The crypto ecosystem is flooded with infrastructure projects. To foster innovation, we should be encouraging more decentralized apps (dApps) and experimentation.

  • When anonymous friends (or "anon frens") attempt to raise funds for their projects, it's crucial to strike a balance that benefits both investors and project teams.

2. Key Principles for Fair Raise

Here are some principles to consider when striving for fair fundraising and project building:

(1) Community-first, Insider Later

  • Avoid early unlocks for insiders to ensure the long-term interests of the project.

    • Examples: The Graph and dYdX followed this principle during their first year.

  • Without proper incentives, project teams may lose the motivation to keep working.

(2) Milestone-based & Allow for Failure

  • Avoid "zombie projects" and soft-rugs by clearly defining whether a project should continue.

    • Funds raised should be placed in a multi-sig wallet and unlocked as needed based on milestones.

  • If a project fails to meet a milestone, let it fail.

    • Responsible teams will face less pressure to grind on without success.

    • Irresponsible teams will need to return the remaining funds.

(3) Embrace Speculation, Not Against It

  • The current trend focuses on utility first and delays token issuance as long as possible.

    • This approach can cause projects to miss time-sensitive opportunities and struggle to get started.

  • Issue tokens as early as possible to:

    • Generate attention via speculation.

    • Invite active participation and contributions, forming a strong community, especially when the valuation is low.

    • Provide incentives to bootstrap the project

  • Bitcoin is the best example for this principle

*These principles mostly applies to dApp teams since they require fewer resources and are easier to adopt than infrastructure projects.

3. Example Terms for a Fair Raise

Here's a hypothetical scenario that both my anonymous friend and I would consider fair and be willing to accept / invest in if the team is competent:

(1) Simple Math

  • $100,000/year cash + $100,000/year tokens for each position (1-year cliff, 3-year vesting; provides enough incentives for top talent)

  • 5 people (for dApp team)

  • $1 million to get started

    • At the start: $10 million valuation (10%+15% booked; 75% reserved for other purpose)

    • By the end of year 1: $20 million valuation, need to raise $1 million through OTC sales

Notes: Mainly for dApp teams; infrastructure projects may need 10x, gaming projects 5x, mass market projects 5x resources.

(2) Execution Details

Before the Raise

  • Whitelist can be used to grant token alloaction

  • Team need to desc Q1 target and supported needed explicitly if raise is successful

After the Raise

  • Multi-sig treasury: cash and tokens.

  • Governance choices

    • Monetary governance vs. non-monetary governance (e.g., product decisions)

      • At the beginning, core team should have more voting tokens in non-monetary governance decision such as product decisions

      • For monetary decisions such as stop project and refund, only non-team token will be accounted.

    • Proposal threshold: ~1% of total (i.e., 1/10 of initial token sold)

  • No liquidity to provide, the only way out is to grow and contribute together.

  • 85% buy-back price during the first month.

  • Quarterly planning milestones & unlock cash and token (if starts vesting).

  • Can issue NFTs instead during the seed round (e.g., BAYC, GAL) if planning a later IDO on a premier exchange.

  • Discount OTC via OHM bond model.

*Not aligned with big spotlight events like Binance IEOs.

(3) Scenarios

(4) Way Out

  • Determine a threshold for continuing if a significant amount of money is returned (e.g., >50%, discontinue and return all funds).

  • If the project fails to raise a second round of funding, consider it a failure.

4. Conclusion: The Path Forward: Encouraging Interesting Projects and Organic Growth

Striking a balance between investor protection and project team incentives is vital in the world of Web3 and cryptocurrencies. By following the principles and example terms discussed in this post, we can foster the development of more interesting projects that grow organically. This approach is preferable to focusing solely on massive raises for infrastructure projects and complaining about insider profits.

By ensuring fair terms for raising funds and building successful projects, we can encourage open discussions and ongoing collaboration, which are key to maintaining a vibrant and innovative crypto ecosystem.

]]> zkzhao@newsletter.paragraph.com (zkzhao.eth) <![CDATA[Crypto Data Landscape 2023.01]]> https://paragraph.com/@zkzhao/crypto-data-landscape-202301 VpLdc8SFq6ZKYBCYclel Mon, 23 Jan 2023 13:17:55 GMT

1. Layers

  • Data App: mostly for users, applications related to data, metrics etc.

    • consumer: targeted at consumer, and into daily use; e.g. @coinmarketcap @coingecko

    • business / pro: targeted at pro users, and generally w/ in-depth query features, i.e., you can query the data in different way you wanted (vs. in consumer app you just get the data. e.g. @nansen , @parsec_finance

  • Data Dev: mostly for developers to dev things related to data

    • analytics: used to analyze data, make dashboard; e.g. dune, flipside

    • indexing: infra used to serve data to front-end users. e.g. thegraph, covalent

  • Data Source: where data come from.

    • on-chain: public blockchains such as BTC, ETH BSC etc.

    • off-chain: CEX, Twitter, Github etc.

the order is also the data flow: data source -> data dev -> data app.

underscore means have decentralized design, e.g. thegraph, ceramic

2. Sector Lists

TBD

personal highlights for new services

  • data app

    • nansen, parsec; wacher, arkham catching up

    • bello, lidar, thirdwave for vertical analytics

  • data dev

    • dune; flipside catching up

    • space and time, zettablock

  • data source

    • ceramic, kwil, polybase

  • community

    • sixdegree, CN community

*should be able to search via twitter

3. Thoughts, Trends and Predictions

  • consolidation begins

  • undisclosed bearish opinion on certain part of the chart, shared w/ my fren in vc and startups fren already

  • infra >> features

]]> zkzhao@newsletter.paragraph.com (zkzhao.eth)