A Silicon Valley professor riding the wave, with a net worth of 18 billion, still stays at the podium and has nurtured 7 AI startups.

Why does this university professor, with a fortune of $2.5 billion (approximately RMB 17.9 billion), still teach in the front - line classroom?

According to a report by Zhidx on September 2nd, recently, after the latest round of financing, the valuation of AI data platform Databricks exceeded $100 billion (approximately RMB 716.2 billion), making it expected to become the fourth most highly valued AI unicorn globally. This has also drawn public attention to its co - founder and chairman, Ion Stoica, a professor at the University of California, Berkeley (hereinafter referred to as UCB), and his legendary experience spanning both the industry and academia.

Stoica, now 60 years old, has been teaching at UCB since 2000 and has never left the campus in the past 25 years. From 2010 to the present, Stoica has been involved in the establishment and management of three important laboratories in the School of Computer Science at UCB. These laboratories have contributed 118 research projects in the era of big data, cloud computing, and AI. Among them, there are many projects with wide - reaching influence and have almost become the core infrastructure in the AI field, such as the big data framework Spark, the distributed execution framework Ray, and the large - model inference framework vLLM.

Ion Stoica (Source: YouTube)

He has also founded or incubated at least 7 well - known startups through personal participation in startups, guiding entrepreneurship, or providing connections and financial support. These startups are distributed in industries such as data infrastructure and generative AI.

In addition to Databricks, whose valuation has exceeded $100 billion, Stoica co - founded the AI managed computing platform Anyscale with a valuation of $1 billion, LMArena (Large Model Arena) with a valuation of $600 million, and the video - stream analysis technology company Conviva with a valuation of $300 million.

Stoica is the doctoral supervisor of Li Haoyuan, the founder of the high - performance AI and data analysis platform Alluxio, and Benjamin Hindman, the founder of the data - center operating system startup Mesosphere. Both startups originated from open - source projects and were guided by Stoica. Currently, Stoica also serves as an advisor to the large - model memory technology startup Letta.

Despite his commercial success, Stoica still cares about education and scientific research. This fall, Stoica will continue to teach in the classroom, teaching undergraduate courses in operating systems and system programming.

Through his connections in Silicon Valley, Stoica has attracted a very luxurious list of sponsors for his laboratory, including well - known companies such as Nvidia, Meta, Ant Group, AMD, Google, Amazon, and Huawei. He also invests the assets he earned from entrepreneurship into scientific research to support the laboratory financially.

When analyzing the reasons for his success to Forbes, Stoica said that it is due to his focus on scientific research: "It is a creative process of constantly exploring new ideas." In this article, we will review Stoica's legendary entrepreneurship and scientific research experiences and understand his thoughts and summaries on his successful journey.

01.

Studied under Zhang Hui, an alumnus of Peking University's School of Computer Science

Started his first startup in 2006

Stoica completed his doctoral studies at Carnegie Mellon University (hereinafter referred to as CMU) in 2000. His supervisor was Zhang Hui, the youngest tenured professor in the history of the school and an alumnus of the Class of 1984 at Peking University's School of Computer Science.

In the same year, he joined UCB and has been teaching there ever since. Located in Silicon Valley, UCB has a strong entrepreneurial atmosphere. In 2006, Stoica launched his first startup project - Conviva.

The founding team of Conviva (Source: Conviva's official website)

In 2006, streaming media content was gradually on the rise, and YouTube had become one of the mainstream platforms. Stoica and his supervisor Zhang Hui saw the market demand for optimizing the online video experience and started their business based on the scientific research achievements of CMU and UCB, co - founding Conviva.

In the early days, Conviva focused on real - time video - stream analysis and became one of the world's earliest platforms to achieve adaptive video playback. The company uses AI technology to identify video - playback quality issues on video platforms and collects user viewing data (including the content being watched, user preferences, etc.) to generate analysis reports.

Conviva's major clients include the US Fox Corporation (FOX) and Peacock, a streaming platform under NBCUniversal.

Stoica originally served as the CTO of Conviva, but he no longer holds an official management position at Conviva. He still remains on the board of directors and meets with the team every week.

Since its establishment, Conviva has completed 7 rounds of financing, with a total financing amount of $110 million. Its last round of financing was completed in 2017, and the post - investment valuation was approximately $300 million.

02.

Building core infrastructure for the big - data era

Transforming achievements into a startup valued at tens of billions

While founding Conviva, Stoica did not stop his academic and scientific research. The School of Computer Science at UCB has a tradition of establishing a collaborative laboratory every five years. Stoica has been deeply involved in the establishment of three major laboratories in the school in the past 15 years, leading the members to achieve results with significant academic and industrial influence.

In 2011, UCB established the AMP (Algorithms, Machines, and People) Laboratory. In the speech announcing the establishment of the laboratory, Stoica said that the AMP Laboratory hopes to understand big data by closely integrating algorithms, machines, and human resources.

During its five - year existence, the AMP Laboratory developed 25 research projects, including the open - source distributed resource management framework Apache Mesos, the open - source distributed big - data processing framework Apache Spark, and the open - source distributed data orchestration platform Alluxio (formerly Tachyon). It has become an important contributor to the field of big - data infrastructure.

These three open - source projects were eventually transformed into startups:

Databricks (total financing of $20.8 billion, valuation exceeding $100 billion)

Databricks is the second startup Stoica participated in founding. Its main business is an AI data analysis platform. Its founding team consists of seven professors and postgraduate students from the University of California, Berkeley, most of whom are from the AMP Laboratory managed by Stoica.

The founding team of Databricks. Stoica is the second from the left (Source: Databricks)

In 2009, the founding team of Databricks developed the open - source distributed big - data processing framework Spark and continuously maintained and updated the project. They even set a world record for data sorting speed.

After operating Spark as an open - source project for some time, Matei Zaharia, an associate professor at UCB and a co - founder of Databricks, said that Stoica hoped to turn Spark into a startup to encourage users to take research from university laboratories more seriously.

In 2013, Databricks was officially established. Relying on the success of the open - source project Spark, Databricks received $14 million in Series A financing from a16z. From 2013 to 2016, Stoica served as the CEO of Databricks and was deeply involved in the company's daily management and decision - making.

In 2016, he handed over the CEO position to Ali Ghodsi, a former visiting scholar at the AMP Laboratory. Regarding this decision, he said in an interview with Forbes: "Staying in this position for longer would mean leaving Berkeley, so I had to make a choice. I chose to go back."

Today, Databricks has grown into a super unicorn with a valuation exceeding $100 billion, serving more than 60% of the Fortune 500 companies. Stoica still serves as the executive chairman of the board of this company.

Stoica serves as the executive chairman of the board of Databricks (Source: Databricks)

Alluxio (total financing of $73 million, valuation not yet disclosed)

Alluxio, formerly known as Tachyon, is a memory - centric, fault - tolerant virtual distributed storage system designed to address the data - sharing challenges within the Apache Spark ecosystem.

In 2015, Alluxio was officially established. Its early investor was a16z, an investment institution closely related to Stoica. Its founder and CEO, Li Haoyuan, completed relevant research at the AMP Laboratory, and his thesis advisors included his doctoral supervisor, Stoica.

Alluxio's current management team. Li Haoyuan is the first from the left in the top row (Source: Alluxio's official website)

As the technological trend shifts towards hybrid - cloud and multi - cloud architectures and AI, Alluxio has continuously developed and provides solutions for data access in AI and data - intensive workloads. Today, Alluxio's AI acceleration platform supports nine out of the world's top ten Internet companies.

Alluxio has completed 4 rounds of financing. Its most recent round of financing was completed at the end of 2021, led by Hillhouse Capital, with a financing amount of $50 million. However, its valuation has not been publicly disclosed yet.

Mesosphere/D2iQ (total financing of $250 million, maximum valuation of $775 million)

Mesosphere was founded in 2013. Stoica's student, Benjamin Hindman, serves as the CIPO (Chief Intellectual Property Officer) of the company. The company's main technology comes from the open - source project Apache Mesos, and Benjamin Hindman is the core author of this project.

DCOS (Data Center Operating System) is Mesosphere's core product. Mesosphere abstracts the data center as a "single large machine" and provides a unified operating platform for large - scale computing and data - intensive workloads through distributed kernel scheduling and resource management.

In 2015, Mesosphere announced that it had received investments from top venture capital firms such as a16z. With the rise of containerization, microservices, and cloud - native trends, Mesosphere gradually transformed its products to support Kubernetes and officially changed its name to D2iQ in 2019. D2iQ's solutions are mainly targeted at hybrid - cloud and enterprise - level Kubernetes platforms, helping customers simplify the deployment and operation of cloud - native applications.

To date, D2iQ has raised nearly $250 million in total financing from investors including a16z, Khosla Ventures, and HP. Its valuation reached $775 million after the Series D financing in 2018. There were rumors that Microsoft, Google, etc. intended to acquire D2iQ, but the deals did not go through due to the opposition of D2iQ's management.

At the end of 2023, D2iQ announced the termination of its operations. The company's assets will be liquidated and distributed to creditors.

03.

Guided the scientific research of the high - performance distributed execution framework Ray

The derived startup is now valued at $1 billion

In early 2017, Stoica participated in the establishment and management of the UCB RISE (Real - time Intelligent Secure Execution) Laboratory. At that time, Stoica and others had realized that the cycle between data generation, computing, and execution was closing, AI had become a reality, and the impact of computing was spreading to every corner of the world.

The RISE Laboratory is in line with the AMP Laboratory. It hopes to further develop relevant technology stacks on the basis of big - data analysis to help applications interact with the environment in a real - time, intelligent, and secure manner.

In the fields of large - scale machine learning and reinforcement learning, the RISE Laboratory has contributed more than 41 open - source projects. Among them, the most influential one is the high - performance distributed execution framework Ray, which has been successfully transformed into the startup Anyscale.

Anyscale (total financing of $259 million, valuation of $1 billion)

In 2019, Anyscale was founded. Its founding team includes Stoica, Philipp Moritz (Stoica's doctoral student), and Robert Nishihara. Stoica currently serves as the executive chairman of the board of Anyscale.

Philipp Moritz and Robert Nishihara are the main authors of Ray. This project has greatly improved the usability of distributed computing, allowing developers to scale single - machine Python programs to large - scale distributed environments with simple APIs. Stoica guided these two authors in completing relevant research on Ray.

A group photo of Stocia and other co - founders of Anyscale (Source: Anyscale's official website)

Ray provides a unified programming method, and developers don't need to worry about the underlying distributed details. Compared with traditional Spark and Hadoop, Ray is more flexible, supporting low - latency task scheduling and various computing modes (batch processing, streaming, online inference, etc.).

This project has a particularly large influence in the AI/ML field. Many popular libraries in the industry are built on Ray, such as the Ray Tune hyperparameter tuning framework and the Ray RLlib reinforcement learning library.

On this basis, Anyscale provides a fully managed computing platform for Ray, adding optimization, observability, data governance, and developer tools, making it one of the best platforms for running Ray workloads.

Currently, developers from leading companies such as Uber