「MIRAISE PREVIEW 2023」Full Version


MIRAISE is the venture capital in Japan focused on startups founded by software engineers. We have invested in 49 software startups based in Japan, US, Estonia, and other countries. (As of December 2022)

Based on ongoing research of software technology and related startups, MIRAISE builds insights through interactions with Japanese and global founders.

MIRAISE is pleased to announce the release of the 2023 report for those who are involved in tech and startups.

This report was known as “MIRAISE TREND”, which has been published annually past years. We have renamed it to “MIRAISE PREVIEW” this year, implying prediction of the future trends of startups that requires an understanding of software technology and a grasp of the signs of major trends. This shouldn’t be called a "trend" based on past events, but rather a forecast and insight based on software technology tied to startups, so we prefer “PREVIEW”.

We would be happy if you could "preview" the unique perspective of the engineer focused Seed VC.

Table of contents

  1. Security demand are being increased in open source software

  2. Serverless evolution comes to databases

  3. Keep an eye on Blockchain Oracles in web3

  4. Investments in Webassembly are increasing

  5. Many Unicorns can be found around Large AI Model

1. Security demand are being increased in open source software

It’s common that developers are building products using a variety of open source software (OSS) today. In January 2022, the U.S.Department of Defense announced that they will prioritize OSS adoption over proprietary products. According to Crunchbase, OSS startups raised a total of $2.8 billion in 2021, double the amount raised the previous year, suggesting that investors are also increasingly focused on OSS.

As OSS proliferates, so do the security challenges of OSS. An Aqua Security survey found that 32% of the top 35 JavaScript package managers "npm" do not use two-factor authentication, leaving accounts unprotected and at risk of account hijacking. A vulnerability in one OSS can cascade to many other software applications that use it and become a major problem in a short period of time. There have also been cases of "man-made disasters" where OSS developers intentionally introduce malicious code.

(Source: GitHub Security Lab has found 372 open source vulnerabilities so far.)

This is also a problem with the nature of OSS. Since most OSS developers are volunteers, they do not earn sufficient income from their development. As a result, in some cases, they may not be able to verify that their development is up to the standard required by companies. In addition, in the area of OSS, authority tends to be concentrated in the hands of a single developer, so we as a community need to quickly discuss how we can solve this problem.

The development of Web software has grown rapidly with the contribution of OSS, however as the use of OSS becomes more popular, it is urgently necessary to solve issues by adopting a security perspective that has shifted from "fundamentally good" to "fundamentally bad ". Some startups have already emerged to monitor OSS code and detect suspicious code. U.S. based Socket has raised $4.6 million in a seed round, and GitHub is also conducting research and development to improve OSS security with its GitHub Security Lab.

The fact that security issues are such a hot topic proves that OSS has become an integral part of product development. In fact, many web3 projects are developed using OSS, and development by the global developer community is creating new innovations. There is no doubt that OSS will continue to play a major role in the Internet world.

2. Serverless evolution comes to databases

Serverless is a system architecture that has been gaining attention in recent years. The most prominent feature is its cost advantage. Since the system is pay-as-you-go for system calls, server costs can be reduced depending on the number of accesses. It also reduces the maintenance costs of keeping applications on servers.

Serverless technology has been used to run relatively small systems and tasks as "serverless runtime," but its use is beginning to spread to databases as well.

In 2022, NEON, which is developing a serverless version of Postgres, raised $54 million, and Upstash, which is developing a serverless version of Redis, raised $1.9 million. This suggests that the development of serverless databases will accelerate in the future.

( Source: SELECT 'Hello, World' - traditional database vs. serverless provided by NEON, separation of storage and computing. )

The reason behind the need for serverless is the shift of databases to the cloud. The three major cloud examples such as AWS, Azure, and Google Cloud Platform are representative of this change. As the database continues to shift to the cloud, trying to get more throughput and storage using a traditional primary/replica configuration database deployment requires a larger cloud and increases costs.

By separating storage and computing, serverless databases can be operated cost-effectively for each. The evolution of Kubernetes container technology is behind the realization of serverless databases. Applying container technology to databases allows for quick scale-up and scale-down.

In many cases, startup companies do not have a sufficient number of infrastructure engineers, and it is difficult to estimate the number of accesses. Under these circumstances, serverless databases are likely to be utilized more and more in the future.

3. Keep an eye on Blockchain Oracles* in web3

* it does not refer to the Oracle database, but to Oracle in web3

Oracles are data feeds from which smart contracts retrieve external off-chain data. Currently, it is mainly used as a price oracle in De-Fi to retrieve external market prices and interest rates. Meanwhile, there has been an increasing number of attacks on De-Fi aimed at Oracle, and in October 2022, Mango Markets, one of Solana's largest De-Fi, was hacked to manipulate prices using an oracle, resulting in $114 million being stolen. This was due to the intentional manipulation of the original data from which Mango Markets’ price oracle is referenced. The oracle is essential for web3 services as it is the link between the real world (off-chain) and the blockchain. Further evolution and improvement will be required in the future.

Here are some examples of web3 services with Oracles.

Chaninlink has created a Decentralized Oracle Network that collects data from across the Internet and multiple external sources, providing not only price oracles but also oracles for sports results, insurance, data from IoT sensors, energy emissions, credentials, and more.
Pyth has established a partner network that provides information and rewards companies that provide excellent data.
Cardinal provides an easy way to prove ownership of the wallet and Twitter account, which is utilized in the operation of DAOs.
Khaos Oracle
Khaos Oracle was developed by Frame00, which is developing a DEV protocol for the creator economy. It is used to monetize creators' assets on the web by placing them on the blockchain through staking and other means.

( Source: Frame00, DEV protocol developer, completes JPY250M in funding - Khaos Oracle connects web2 and web3. )

The reason why Oracle is attracting attention is that the scope of web3 applications is expanding in many directions. For example, NFT is often associated with "ownership of digital art", but now it has an infinite number of uses, including items for starting games, community and real estate rights, proof of participation in events, and a means for individuals to manage their behavior and purchase history. Oracle is essential to bridge the gap between web3 applications and the real world, and will continue to grow in importance in the future.


The etymology of the word Oracle, based on the metaphor that the on-chain is God's kingdom and the off-chain is the world we live in, the name seems to be a connection between the two. It is a gateway for the transmission of otherworldly information.

4. Investments in Webassembly are increasing

Webassembly (Wasm) is a technology that bridges the gap between "various programming languages'' and "many different execution environments". Code written in more than 30 different languages, including C, C++, C#, Rust, Go, Swift and others, can be compiled into .wasm files that can be run in browsers, on servers, or on edge devices.  

Currently, in order to run a program in a browser, it must be written in JavaScript. JavaScript was born in 1995, and thanks to browser competition, its execution speed has increased. And JavaScript has also moved beyond the browser to be used on the server side (Node.js). On the other hand, as content on the browser becomes richer, such as in streaming services and VR, AI and other computations explode and further acceleration is required. Wasm is the technology to meet this challenge.

Amazon Prime Video has adopted Wasm and announced on their blog that it is 10 to 25 times faster than JavaScript. The Polkadot network, which allows different chains in the blockchain field to interoperate, also uses Wasm to absorb the language and runtime of each chain. Even Wasm, which seems like a panacea, still has its challenges. Debugging and build tools are not as comprehensive as they could be, and support for non-browser APIs is still needed.

( Source: The State of WebAssembly 2022 - a survey of developers on their intended use of Wasm, which is used in a wide range of development settings, including web, IoT and Blockchain. )

Solomon Hykes, founder of Docker, says: "If WASM+WASI existed in 2008, we wouldn't have needed to create Docker. That's how important it is. Webassembly on the server is the future of computing. (on Twitter)", he highly praised. The magical Wasm, which is independent of the execution environment and allows developers to write code in the language of their choice, is highly anticipated by investors, and in 2022, Fermyon, which provides a cloud for Wasm, raised 20 million USD, and Cosmonic, which provides Paas for Wasm, raised 8.5 million USD. More investment capital may flow to Wasm related startups in the future.


Historically, the excitement of Virtual Machines (VM) at the height of their popularity comes clearly to mind.. Many experienced developers may recall Sun Microsystems' "Write once, run anywhere (WORA) ” when Java was first introduced. It might be fair to say that Wasm applies this to the language side as well.

5. Many Unicorns can be found around Large AI Model

Large AI models are fundamental models of AI trained on generic datasets with over billions of parameters. AI models are developed by well-funded AI research organizations such as Google, Microsoft, and OpenAI because they require large amounts of computational resources for training. Until now, Google and Microsoft have not released their models to the public for fear of abuse. In 2022, Stability AI in the UK released a similar model, Stable Diffusion, as open source. This makes the Large AI model available to general developers.

( Source: GitHub - Stability-AI/stablediffusion - Stable Diffusion is open source and publicly available.)

A common approach to AI development in recent years has been a pre-training/fine-tuning approach, in which neural networks are trained on large generic datasets using large amounts of computational resources, and then fine-tuned for the task using less data. By using large AI models for this pre-training, startups can build products with far less data and computational resources than if they had to build models from scratch.

GitHub Copilot utilizes OpenAI's Large AI model, GPT-3, which 'translates text into different languages', as its programming language, allowing developers to simply write 'code for process A' in colloquial language and the program is automatically generated, helping developers to be more productive. The company has been a great success so far.

The most successful startups to date have not been those that took advantage of technological advancements, but those that used them to provide solutions to the problems in ways that were not possible before. Think of Google and Facebook taking advantage of technological advances in the internet, Uber and Instagram taking advantage of technological advances in mobile, Ethereum and OpenSea taking advantage of blockchain, etc. Leveraging the Large AI model has the potential to create new unicorn companies.

MIRAISE offers information

In addition to this report, MIRAISE provides weekly updates on the latest tech news, so please check here.

To close

Thank you for reading this report.

If you are a new business department or investment department who would like to discuss with MIRAISE after reading this report, or if you are an engineer who is thinking of starting a business and would like to discuss with MIRAISE, please get in touch with us from here.

Inquiry form