Explaining Starkware in detail: underlying design, Cairo language, team and economic model

At present, the development trend of Ethereum is more and more inclined to modular block chain.

Under the architecture of modular blockchain, the transaction calculation and execution on Ethernet is no longer operated by the main network. This part of the work is transferred to Layer 2 under the chain to complete, which reduces the burden on the main network, greatly expanded the performance of Ethereum.

In the future, Ethereum’s main network will only be responsible for consensus and data availability, while Layer 2, which is responsible for all computing execution, will become a cloud-like service provider, like the Amazon Cloud and Alibaba Cloud.

In the case of Layer 2’s scaling scheme, Rollup is considered the most promising.

Rollup verifies and computes the transactions under the Ethernet Square chain, packs them up, compresses them, and sends them back to the main network with the new State root after the transaction update, thus reducing the footprint on the main network space.


There are two further implementations in Rollup, Optimistic Rollup and zk-Rollup. Among them:

  • Optimistic Rollup adoption First, assume that all transactions uploaded to the main chain are legal, and set a challenge period, allowing the verifier to challenge the questionable transaction with a certificate, and punish the fraudster once the fraud is proved to have occurred, reward the challenger
  • Zk-rollup uses All transactions are verified and packaged under the chain, and the verified transactions are submitted to the main chain with zero-knowledge proof to prove the validity of the transaction. In the words of StarkWare CEO URI: “It provides untrusted computational integrity, the ability to make sure — even without oversight — that the computation is being performed correctly.” Much like the idea of early bitcoin

There are three main roles in Rollup:

  • Sequencer: Is responsible for collecting transactions from users, sorting them and passing the new Merkle Root to Layer1’s rollup smart contract, similar to miners
  • Prover: is responsible for calculating and validating all transactions and generating a zk-proof to verify the validity of the transaction
  • Verifier: is typically deployed on Layer 1 and is responsible for doing some of the calculations to verify the validity of Prover’s proof submission to ensure that it provides all the information required for computational integrity

Optimistic Rollup is EVM-compatible, technology solutions mature and arrive early, the cost of migration is low for developers, and represents the projects Arbitrum and Optimism with the highest market share in Rollup. Because zk-rollup is not compatible with EVM, it is difficult to develop, and its application is not as wide as the intelligent contract oriented OP ROLLUP.


However, zk-Rollup has a number of advantages over Optimistic Rollup, such as:

  • Extensibility is better: This is due to the fact that zk-Rollups requires less data to be uploaded to the main network than Optimistic Rollups. Because all the trading information posted to the chain by zk-rollups has been verified and attached to the certificate, others only need to verify the certificate and do not need to verify the transaction, so the data related to the verification transaction does not need to be uploaded, optimistic Rollups, however, needs to link this data as well because of the challenge of proof of fraud. In practice, zk-Rollup is about ten times more capable of improving performance than Optimistic Rollups
  • The transaction was finalized within a short period of time: Transactions posted to the chain by zk-Rollups have been validated and accompanied by proofs, which are relatively simple and usually take about 10 minutes to validate, but transactions submitted by Optimistic Rollups are subject to fraud risk, the challenge of proof of fraud, which has a time window of about a week, does not allow for instant withdrawals
  • Higher security: The transactions uploaded by zk-Rollups are verified and bear witness to the low probability of malfeasance. Zk-Rollups has better data reliability and security than Optimistic Rollups, which may upload transactions with fraudulent content

Vitalik has 2021 that “Optimistic rollups will win in the short term because of their EVM compatibility, and in the medium to long term, as zk-SNARK improves, zk-rollups will win all the use cases.”, vitalik is more bullish on zk-rollups.


The two leading ZK projects are zksync and Starkware. ZKSYNC uses Zk-snark technology, while Starkware uses zk-STARK technology.

zk-SNARK is the abbreviation of concise non-interactive proof, zk-STARK is the abbreviation of Extensible Transparent Knowledge Proof. Zk-snark birth time is early, the technology is relatively mature, at present has been a relatively wide range of applications. Zcash, an early privacy money program, used zk-SNARK technology. zk-STARK is considered to be an improved version of zk-SNARK, a faster and more convenient way to implement zk-SNARK, designed to optimize zk-SNARK to make up for some of its shortcomings.

As the name suggests, zk-STARK is scalable and transparent compared to zk-SNARK, and zk-STARK offers greater security. Specifically:

  • Extensibility: With the help of zk-STARK technique, the operation time of the certifier is at most a quasilinear of the scale of computation and the verification time is a polynomial logarithm of the amount of computation. In zk-SNARK, the certifier may face a complicated and expensive proof process, and the corresponding longer verification time
  • Transparency: zk-STARK does not require trusted initialization settings, unlike zk-SNARK, which requires a trusted public reference string, but instead uses random public validation, so there is no fear of parameter corruption or modification;
  • Security: Since zk-STARK relies on hash function for symmetric encryption, the collision-resistant nature of Hasche’s function provides protection against quantum attacks

Zk-snark is still the mainstream in ZK field, but zk-STRAK technology is difficult and lack of developers, so it has not been widely adopted. But with significant advantages, we believe that in the future zk-STARK will be the more acceptable solution.


Starkware co-founder Eli Ben-sasson is one of the world’s top cryptographic scientists, a Technion – Israel Institute of Technology Professor, and the co-inventor of zk-SNARK and zk-STARK Technologies. The project was developed in Turing’s full-fledged Cairo language.

How it works

The workflow for the STARKWARE product line is shown in the following figure:


In the case of StarkEx, the workflow of StarkNet is roughly similar.

  1. The user first performs an action in the application, usually the application’s client or web page front end, and the application performs these actions and sends the transaction to the STARKEX service;
  2. The STARKEX service performs these transactions in packages and sends the packages to The execution of this process is done by a program written in the Cairo language, and it’s worth noting that Starkware uses Cairo, the Turing-complete language, the computation results of all intelligent contracts are transformed into provable polynomial equations, so that the intelligent contracts are compatible with the validity proof
  3. To generate a certificate that can prove the validity of the batch transaction;
  4. Sharp sent the Stark certificates to the deployment Verifier does the verification;
  5. StarkEx Service sends a status update transaction up the chain to , the smart contract will accept this new state change only after Verifier completes the proof


STARKWARE currently has two products:

  • For applications Applications running on StarkEx can customize their business logic. StarkEx serves applications such as DEFI and games to speed up transactions and reduce transaction costs, as well as data availability through either ZK-Rollup or Validium
  • Is a license-free decentralized, anti-censorship Any user and developer can deploy a smart contract written in the Cario language

The two products correspond to two different business models, the first offering a customized service for the application and the second selling it, the latter provides a generic service without a license and then charges a transaction fee on the network or acts as a node to extract mevs on a cost-first basis.

Eli Sabasson, co-founder of StarkWare, once compared StarkEx and StarkNet to computers and the cloud, respectively. Buying StarkEx is like having your own custom-built computer, and each customer has their own scaled-up computer. StarkNet, on the other hand, is a network, a universal service that everyone can pay for and use.

The working process of StarkNet is shown in the following figure. The big difference is that in StarkNet, Sequencer replaced StarkEx Service to validate transactions, package transactions, call the Cairo program, and SHARP to generate certificates, prover is responsible in StarkNet. Currently, both Sequencer and Prover are designated by StarkWare’s official team, raising questions about centralization.



StarkNet has posted a road map on their blog and says they are moving toward decentralization.


In this scenario, the team-building process for StarkNet consists of four phases:

  • Step 0-infrastructure
  • Step 1–Planets: Rollup of a single operator, single app
  • Step 2 — Constellations: Rollup of a single operator, multiple apps
  • Step 3 — universe: decentralized operator, Rollup of multiple apps

In Step 1, with only one application running on a StarkNet instance (a rollup) , the developers focused on how to implement their business logic on a StarkNet-enabled network.

The next step will support running multiple applications on the same StarkNet instance, and these applications will share the same global Layer 2 state. This will allow interoperability between different applications and lower gas costs due to economies of scale. At this stage, StarkNet will become a fully functional framework for running multiple applications with arbitrary business logic on top of ether, with each instance run by a single operator. From the user’s point of view, StarkNet works much like Ethereum, only more extensible.

The final stage is the stage where StarkNet decentralizes the operator. In this phase, the team focuses on two things: (1) using Rollup to achieve a better consensus mechanism and (2) using economic models to motivate contributors and operators, attract more participants to decentralize the network.

The Cairo language

Cairo (CPU Algebraic Intermediate Representation) is a turing-complete programming language used by Starkware to generate STARK proofs for general computations. Application developers can use Cairo to define any business logic and perform both under-and up-chain validation without having to manually generate complex circuits or AIR (algebraic mediations) .

Through the Cario programming language, any computing proposition can be fast and safe coding proof. A Cario application can prove its execution logic, and a Cario application can integrate multiple Cario applications into SHARP.

Sharp (shared certifier) can aggregate transactions for several separate Cario applications and use them in the same STARK certifier. Using the STARK proof, the time required to present the proof is approximately linear to the time required to execute the statement. The time required for verification is logarithmic to the time required to present the proof.

Since the verification time is much shorter than the verification time, recursive verification can be implemented: A Cairo program can verify the correctness of multiple transactions, as well as writing a Cairo program to verify multiple Stark proofs. Since only logarithmic time is required to implement validation, such recursive validation can greatly reduce the validation time.

The Starkware team launched Cairo’s Generic Proof Service (GPS) , which developers can use to build their applications in the Cairo language. They can send their Cairo code containing business logic to GPS to prove it, and verify them on the chain.

Since GPS supports the use of a certificate to assert the integrity of execution of multiple different applications, these applications can share the gas cost of the certificate verification.

Team and financing

Starkware’s israeli-based team, consisting mainly of computer science and cryptography developers with a background in Hebrew University of Jerusalem and Israeli technology, is arguably the strongest team in the zero-knowledge proof field.

According to Starkware’s website, the team currently has 77 members, plus seven subject consultants and five other consultants. Key members include:

  • Eli Ben-sasson is a Co-Founder & Chief Scientist with a phd in Hebrew University of Jerusalem computer science and a long history of research in computational integrity, cryptography, and zero-knowledge proof. Professor of Technion – Israel Institute of Technology Computer Science. Founder scientist of Zcash, inventor of zkSNARK and zkSTARK
  • Uri Kolodny: Co-Founder & CEO, Hebrew University of Jerusalem Bachelor of Computer Science, is an experienced, collaborative serial entrepreneur
  • Alessandro Chiesa: Co-founder & Chief Scientist, Professor of Computer Science at the University of California, Berkeley. Zcash’s founding scientist, co-inventor of zk-SNARKs, and core developer of libsnark, a leading open source library for concise zero-knowledge Proof
  • Michael Riabzev: Co-founder & Chief Architect. Ph. D. , Israel Institute of Technology, worked for Intel, IBM
  • Oren Katz: Vice President of Engineering. A Hebrew University of Jerusalem computer science graduate with a Tel Aviv University MBA and 20 years experience as a senior engineer

In terms of financing, StarkWare disclosed seven rounds of financing, involving 45 investments, and disclosed a total of $273 million. In 2022, Series D was valued at $7.9 billion, with investors including Paradigm, institutions such as Sequoia Capital and Pantera Capital and Vitalik.

Economic models

Starkware published their tokens economic model in a blog post in mid-july. It was originally scheduled to be issued in September, but news of the delay was later released. As of the time of publication of this article, there is still no official issuance of the coin.

The StarkNet token will be used in three main scenarios: pay for the StarkNet network, pledge to participate in the network consensus, and vote on community governance.

  • Currently, costs in StarkNet are paid in ETH. But next, the project is expected to be paid using only native StarkNet tokens. To support a good user experience, automated and decentralized mechanisms on the chain will allow users to pay with ETH
  • Certain services that are critical to StarkNet’s viability and security may require a pledge of StarkNet tokens. These services may include sequencing, an interim L2 consensus before L1 finality is reached, Stark certification services, and data availability provisioning. These services are expected to be decentralised in the 2023
  • The proposal to improve StarkNet will require a minimum token holding threshold. All changes to protocols that are critical to StarkNet’s viability, security, and maintenance require a vote, either directly or through delegation. For example, all major updates to the StarkNet operating system require the approval of the token holder

The initial circulation of the StarkNet tokens was 10 billion, with 17 per cent allocated to StarkWare investors, 32.9 per cent to core contributors and 50.1 per cent to the foundation. At the same time, the token has an add-on mechanism, and a portion of the new coinage and transaction costs will be awarded to core infrastructure developers and smart contract developers.


51 per cent of the specific uses allocated to the foundation are:

  • 9% — eco-developers, users, and communities. This includes some users who interacted with StarkEx prior to June 1
  • 9% — Rebates — refunds of StarkNet tokens to partially cover the cost of joining StarkNet from Ethereum
  • 12% — funds researchers and workers who develop, test, deploy, and maintain the StarkNet protocol
  • 10%-strategic reserve to finance ecosystem activities consistent with the foundation’s mission
  • 2% — donations to prominent institutions and organizations in related fields, such as universities, non-governmental organization, etc. , as determined by the holders and foundations of StarkNet tokens
  • 8.1% unallocated-to further support the STARKNET community in a manner determined by the community

Taking into account the customary allocation of decentralized ecosystems and the long-term interests of communities, 49.9 per cent of investors and core contributors will have a four-year lock-in period, there will be a 1-year Cliff and a linear release of tokens.

As can be seen from token allocations, StarkNet tokens reward developers who prioritize core infrastructure and DAPP over the average user.

Operating data

Due to its early launch and the relative maturity of the Scale as a service model, StarkEx Tob has accumulated a number of users such as dYdX (which has since left) , Immutable X, rhino.fi , Celer and Sorare. According to the company’s website, the cumulative value of transactions on StarkEx is $732 billion, compared with $461 million for TVL.


Another product, StarkNet, currently has more than 100 apps in the ecosystem, according to www.StarkNet-ecosystem.com, with the largest number of DEFI (45) , NFT (24) and infrastructure (18) .


Among the more prominent projects, both AAVE and Maker will be deployed to StarkNet, two days after AAVE announced the completion of the first phase of its cross-chain partnership with StarkWare. This phase is the AAVE v 2 aToken cross-chain to StarkNet smart contract infrastructure on Ethernet, allowing network users to get cross-chain aToken at StarkNet. As early as June, Maker had launched Dai’s transfers between the main network and StarkNet.

It is worth mentioning that DYDX, which previously occupied the majority of TVL in StarkEx, left in June, choosing a separate blockchain based on the Cosmos SDK to launch the V4 version of the product. This incident has somewhat dented investor confidence in layer 2 and Starkware.

The exodus also shows that applications with real users and influence have more say and initiative in choosing the underlying infrastructure, which does not depend on a single underlying chain, you can actively choose other low-level public chain, or even build their own requirements in line with their own application chain.

The old“Fat protocol thin application” narrative may be redefined. For all kinds of low-level infrastructure projects, how to seize the quality of their own ecological applications, to do a good job of preservation, has become the current need to rethink the problem, after all, to capture more value, you need to capture more users, and it is the application, not the underlying technology, that really interacts with users. Users don’t care how the underlying technology is implemented, they just migrate as the application migrates.


Starkware, with a strong technical team, is working on the product step by step according to their plan: on October 26th, they announced that they had written ZK-EVM using the STARK proof programming language Cairo.

At the same time, they also face some challenges. For example, zk-STARK currently has high technical difficulty, slow development speed and few developers, which brings many difficulties to the development and retention of ecological applications. DYDX’s departure took away most of the TVL in the project, exposing the low-level infrastructure to the high-quality application binding capacity of the predicament, but also let people see the high-quality application in the selection of the low-level chain strong voice.

However, due to the advantages of the technology, in the long term, we are still optimistic about the development of zk-rollup and Stark. Later, Starkware will continue to be one of Rollup’s strongest contenders as the tokens are released and the network moves toward decentralization.

Like what you're reading? Subscribe to our top stories.

We will continue to update Gambling Chain; if you have any questions or suggestions, please contact us!

Follow us on Twitter, Facebook, YouTube, and TikTok.


Was this article helpful?

93 out of 132 found this helpful

Gambling Chain Logo
Digital Asset Investment
Real world, Metaverse and Network.
Build Daos that bring Decentralized finance to more and more persons Who love Web3.
Website and other Media Daos

Products used

GC Wallet

Send targeted currencies to the right people at the right time.