This article is written on the basis of public information, only for information exchange, and does not constitute any investment advice.
Once the wheel of fate starts to turn, it can never turn back.
At the end of 1990s, with the emergence of a technology called Internet in the field of civil commerce, the gears of fate began to turn, and several small companies in Silicon Valley of the United States-Google, Amazon and Facebook-were established one after another until they became the largest technology companies in the world in the future. Correspondingly, China’s Baidu, Alibaba and Tencent have become the other poles of the Eastern Hemisphere.
In 2023, the wheel of fate began to turn again. This time, it’s AI.
In the first year of this AI model, the smoke of "Hundred Models War" not only permeated China. Europe and America are also the same. Those AI native unicorns whose names we can’t name are all vying for a ticket, dreaming of becoming candidates for the next generation of "FAANG" in the next decade or so.
High-end dreams only need the simplest way of introduction. In this article, we will sketch 19 European and American AI unicorns in the big model era. Maybe one day, some of these names will be heard like thunder-just like Google Amazon and Baidu Alibaba more than 20 years ago.
StabilityAI is probably the hottest AI big model company besides OpenAI, and it is also one of the most exposed AI unicorns in Europe and America in China.
Headquartered in London, England, StabilityAI is the company behind StableDiffusion. EmadMostaque, the founder and CEO, is 39 years old and comes from Bangladesh and grew up in the UK.
StableDiffusion developed by StabilityAI has become the most popular text-to-image generator in science and technology. It has been trained by 4,000 A100 graphics cards, and has been downloaded and authorized by more than 200,000 developers around the world. It is also one of the most available open source image models.
HuggingFace is probably the most popular open source machine learning community and platform in the world.
HuggingFace, which has experienced several rounds of financing, has become an AI unicorn with a valuation of $2 billion. Its 24 investors include LuxCapital and Sequoia Capital.
When it first appeared five years ago, HuggingFace was just an iPhone chat bot application for teenagers. The moment of its viral spread appeared in 2018. HuggingFace opened a Transformers library on github. Although the chat bot business was not started, their library quickly became a fire in the machine learning community.
At present, it has shared more than 100,000 pre-training models and 10,000 data sets, and has become the github of machine learning, setting a record for the fastest growing AI project in GitHub’s history. More than 10,000 organizations have used it as a starting point to create AI-driven tools for their business, including more than 1,000 paying users.
Viz.ai was established in 2016 and is headquartered in San Francisco, USA. Viz.ai uses deep learning algorithm to automatically analyze CT neural images to detect the indexes related to stroke and provide clinical decision support.
Viz.ai’s iconic products are VizLVO and VizCTP platforms, both of which have been approved by FDA and listed in the United States.
VizLVO uses artificial intelligence to automatically detect large vessel occlusion (LVO) in patients with suspected stroke, and the notice of suspected LVO is sent to neurologists and interventional doctors, who can view CTA images of patients on mobile devices. The time between the picture and the notification is less than 6 minutes. Stroke experts can initiate the transfer of patients to the intervention center through the existing emergency call function.
Another platform, VizCTP, uses advanced processing technology to automatically analyze CT perfusion neuroimages and generate parametric perfusion color images. According to the recent AHA/ASA guidelines, the treatment window of mechanical thrombectomy was extended from 6 hours to 24 hours, which brought a new suitable treatment cycle for stroke patients.
The progress of artificial intelligence and deep learning makes it possible to analyze medical imaging quickly and accurately. ALADIN multicenter trial proved the ability of VizLVO to detect great vessel occlusion (LVO) in patients with suspected stroke retrospectively, accurately, consistently and quickly on 875CTA images. The left ventricular flow measured by VizLVO extends from ICA-T to MCA-M1, with a sensitivity of 90% and a specificity of 86%. Therefore, VizLVO is considered to be the most accurate LVO detection algorithm currently available in the United States.
Recently, Viz.ai announced the completion of the $100 million Series D financing, which was led by TigerGlobal with a valuation of $1.2 billion.
DeepL is a deep learning company in Germany that develops artificial intelligence for languages. Founded in 2009, it was originally named Linguee.
In 2017, the company launched the DeepL translator based on machine translation system, which has become one of the most popular translation software at present. Developers say that they use a new neural network structure that has been greatly improved, which makes DeepL’s translation more natural than similar services.
At present, more than 1 billion people have used DeepL’s services. By the end of 2021, DeepL was the ninth most visited website in Switzerland, the 30th most visited website in Japan and the 16th most visited website in the world. DeepL is used by 20,000 enterprises around the world.
In January, 2023, DeepL completed the financing of $100 million led by IVP, and the company’s valuation reached $1 billion.
Founded in 2014, Optibus is an end-to-end cloud native SaaS platform supported by artificial intelligence and optimization algorithms to simplify transportation planning, scheduling, scheduling and operation.
The company’s platform has been adopted by transport agencies and operators in more than 1,000 cities in Europe and America, including StagecoachGroup, the largest public transport operator in the UK.
Optibus is used to track and support about 2.5 billion bus, tram, light rail and subway trips around the world every year.
Optibus integrates data about public bicycles, scooters and taxi access points, and currently cooperates with public transport providers in more than 450 cities around the world, including London, new york, Los Angeles, Melbourne, Hong Kong and Brasilia.
Vesttoo is a risk management platform headquartered in Tel Aviv, Israel, which provides insurance-linked investments for all types of asset management companies, while enhancing risk transfer and liquidity in the life and property insurance market.
Vesttoo connects the insurance industry and capital market with artificial intelligence-driven technology and expertise in data science, insurance and finance. Relying on machine learning (ML) technology, Vesttoo creates insurance-related assets from low-volatility insurance liabilities, providing insurance companies with additional sources of funds and providing investors with low-volatility assets.
Quantexa is a large start-up company headquartered in London, which pioneered a method based on machine learning to search company data for the financial services industry, looking for signs of fraud, money laundering and other illegal activities.
After starting from financial services, Quantexa expanded its business to other areas facing financial crimes, such as health care, government and insurance, in order to apply its decision-making intelligence to more scenarios, such as building a complete customer profile and understanding customer compliance. In government departments, Quantexa’s software is also used to identify illegal activities such as human trafficking.
At present, there are hundreds of customers from about 70 countries using Quantexa software, including large enterprises such as Bank of new york (BNY Mellon), HSBC, Standard Chartered Bank, Danish Bank, Vodafone and the Public Sector Fraud Authority under the Cabinet Office of the United Kingdom.
Dataiku was founded in Paris and moved to new york at the end of 2016. It is a leading daily artificial intelligence platform, which provides "daily artificial intelligence" by "opening data and enhancing cooperation within the organization".
During 2022, the company’s customer base increased to more than 500, including more than 150 largest enterprises in the world.
Dataiku is said to be able to apply artificial intelligence technology to various scenarios, including predictive maintenance, supply chain optimization, precision engineering quality control, marketing optimization, etc., involving banking, retail, manufacturing, pharmaceuticals, public sector and other fields, with an annual recurrent income (ARR) of more than 150 million US dollars.
Lightricks Company was founded in Jerusalem, Israel in 2013. Its four main technicians have worked in Microsoft, Google, Adobe and other companies, and their main research interests are computer vision and graphic processing.
The founding team launched a series of image and video editing tools for social media users, such as Facetune2, Facetune Video and Filtertune.
In addition, the ecosystem built by this company also has products with wider design uses, including Boosted, Beatleap, Photoleap, Videoleap and Motionleap. These applications currently have 547 million downloads and installations, and have 29 million monthly active users, most of whom are content creators.
Anthropic was founded in 2021, and the company was founded by former members of OpenAI.
According to Anthropic official website, their target customers are corporate customers, focusing on the research of AI security and reliability. The business model is to provide AI research and development services to corporate customers and sell products and services that give priority to safety.
The core product of Anthropic is "Meet Claude", which is the next generation of artificial intelligence assistants supporting tasks of all sizes. Anthropic’s core products and services include AI research and development services. Their AI research and development services include developing customized AI systems for corporate customers, and conducting AI safety and reliability research. Priority is given to safe products and services including explanatory, controllable and safety testing tools.
These tools are designed to help customers ensure that their AI systems are safe and reliable and can meet regulatory requirements. Anthropic’s focus on safety and reliability distinguishes it from other AI research and development companies and makes it an attractive choice for customers looking for AI systems that give priority to safety and reliability.
Founded in 2019 and headquartered in Toronto, Canada, Cohere is a cloud hosting application provider focusing on generative artificial intelligence models.
Cohere’s artificial intelligence platform is specially designed for enterprises, focusing on providing enterprise customers with AI capabilities equal to large models in text search, text classification and text generation.
In the publicity, Cohere claimed that the characteristics of their products are high performance, high security, cloudy adaptability and data controllability. These characteristics just solve the general concerns of commercial customers about AI at present, and are also the key to their differentiated competition with OpenAI.
Runway is an American AI video software company. Founded in 2018, it has provided technical support for the special effects of Instant Universe, participated in the development of Stable Diffusion, and released an AI video generation tool Gen known as "Midjourney in the video industry".
Recently, Runway released two generations of video generation models, Gen1 and Gen2, using the latest advances in computer graphics and machine learning. GEN 1 also needs to provide the original source frequency, while Gen2 only needs a few words to generate short videos.
Jasper was founded in 2021 and headquartered in Austin, Texas, USA, focusing on the field of artificial intelligence-driven copywriting software.
Jasper’s core product is a generative artificial intelligence platform designed to simplify content creation. It can handle tasks such as updating social media, writing blog articles and creating advertising copy, so that users can focus on more strategic tasks. The platform also supports customization to better meet the unique needs of enterprises, so that they can maintain their brand voice and cooperate effectively.
In April 2023, Jasper has accumulated about 100,000 paying customers, including well-known companies such as Deloitte and Canva, and raised $143 million, with a valuation of $1.5 billion.
Founded in 2016, Replit, headquartered in San Francisco, USA, is one of the fastest growing coding platforms in the world. Ghostwriter, a collaboration tool developed by Ghost Writer, allows developers to collaborate on code processing in real time, and currently has 20 million users.
Replit provides an online collaboration IDE that supports multiple programming languages, including JavaScript, Python, Go and C++.
With Replit, users can share a workspace with one or more users, view real-time editing across files, send messages to each other and debug code together. In addition, users can share projects, ask for help, learn from tutorials and use templates.
Influence was founded in 2022 and headquartered in California, USA. It focuses on developing consumer-oriented artificial intelligence products and is considered as the main competitor of OpenAI.
The core product of Influence AI is a personal AI named Pi, which is a brand-new kind of AI. It is designed as an understanding partner, providing dialogue, friendly suggestions and concise information in a natural and fluent style.
Pi’s experience is designed to take precedence over human conversation, while other AI’s serve productivity, search or answer questions. Pi can be a coach, a friend, a creative partner or an object of "venting emotions".
Founded in 2022, Adept is headquartered in San Francisco, California, USA, focusing on the development of artificial intelligence models. Adept is said to be building a brand-new machine learning model, which can interact with everything on the computer to change the way people use computers.
Adept’s main product is ACT-1 model, which is a large-scale model for action. It has cutting-edge ability, can receive complex user requests of a series of software tools and execute them, and even coordinate across multiple tools when necessary.
Adept’s competitive advantage mainly comes from its innovative AI technology and strong founding team, including the inventor of the Transformer model. In addition, their ACT-1 model shows certain commercialization potential, which can handle and execute complex user requests of a series of software tools.
Founded in 2021, Character.Al is headquartered in Palo Alto, California, USA, focusing on the application field of chat robots.
The core product and service of Character.Al is a neuro-linguistic model chat robot Web application, which allows users to create and customize chat robot "characters" with specific personalities and parameters. These chat bots can be used for various purposes, including creative writing, text-based adventure games and language learning.
Character.AI claims that since the C-port was launched six months ago, Character.AI has enabled users to create customized AI partners with specific personalities and values, and the monthly website visits have reached 100 million times.
Glean, founded in 2019 and headquartered in new york, USA, is an enterprise-level AI search service company.
Glean is said to have been connected with 100+ SaaS applications, and users can search enterprise data across applications and generate personalized answers and results for different users.
Glean combines vector search and keyword search to train customized AI models for each client company according to its language and background, and constructs a knowledge map about the company’s personnel, content and interaction.
Each user’s search results will be personalized based on the knowledge map. In addition, users can also perform lightweight functions of SaaS applications connected to Glean, such as directly starting a meeting on Glean, creating Jira documents and so on.
This article comes from WeChat official account’s "Brocade" on WeChat (ID: jinduan006). Author: Qi Xiangyang, published by 36Kr with authorization.