The large-scale models represented by ChatGPT and GPT have already formed a tsunami effect in China, and almost all top technology companies are trying to find ways to enter the large-scale model track. The greatest value of large models lies in the general improvement of personal productivity, and companies from all walks of life are actively looking for opportunities to apply large models and generative AI to generally improve the productivity of all employees.
At the SAS Innovate conference held on May 9, 2023, Jim Goodnight, the founder and CEO of SAS, known as the king of global data analysis, said that the large model did show intelligence beyond expectations, although the large model Full sentences and articles can be created, but big models don't express facts all the time, and the risks of big models can't be ignored.
(SAS founder and CEO Jim Goodnight)
SAS did not choose to follow the big model. At the 2023 SAS Innovate Conference, SAS announced that it will invest US$1 billion, or nearly RMB 7 billion, in the field of industrial AI solutions in the next three years. These investments will significantly improve the AI productivity of enterprises in various industries. What is Enterprise AI Productivity? A test report released at this conference shows that SAS Viya's AI and analysis speed is 30 times faster than similar commercial and open source products, the cost is reduced by up to 86%, and it shows stronger scalability!
Analytical calculation engine far superior to competitors
SAS is a world-renowned leader in advanced and predictive analytics software with a history of nearly 50 years. The core technology of SAS is the computing engine that supports various advanced analysis, predictive analysis and AI analysis, from the earliest mainframe to the later minicomputer and PC, from single-machine single-thread computing to multi-core high-performance distributed computing architecture, and then to Today's cloud-native computing engine, SAS has been committed to improving the speed and performance of advanced analysis, predictive analysis and AI analysis, and continuously reducing costs.
Today, SAS Viya is a new generation of cloud-native large-scale parallel computing engine of SAS. Since its launch in 2019 at a cost of US$1 billion in research and development, it has undergone continuous development and improvement, and has won various market research reports. In order to further demonstrate the performance of SAS Viya, SAS cooperated with The Futurum Group, a market research and technical testing company, to conduct comparative tests between SAS Viya and several mainstream commercial and open source AI/ML software products in the market.
This comparison involves more than 1,500 tests, all of which are based on the same computing instance on Azure Cloud, in order to compare the results more accurately - SAS Viya is 30 times faster than competitors on average, faster than H2O and SparkML, Also faster than an "A" competitor; SAS Viya scales better than the competitor, especially performing better on large datasets; the best result achieved by SAS Viya is 326 times faster than a competitor , on average 49 times faster.
The Futurum Group found that the larger the data set, the better SAS Viya performed. In particular, when training a model on a 300 million dataset, SAS Viya took less than 12 minutes to train, while SparkML and an "A" competitor failed after running for hours - failing to produce any result. In fact, SAS Viya not only outperforms all competitors, but is the only software platform that is continuously scalable.
The Futurum Group believes that SAS Viya can deliver key business decisions faster, which in turn means lower costs. Due to the accuracy and speed of SAS Viya, business users can have more model choices. That is, enterprise data scientists can try more models and choose the one that produces better results without increasing their budget.
Russ Fellows, an analyst at The Futurum Group, said that he has been in the IT industry for 30 years and has worked as an analyst for 15 years. It is good that other tests can achieve an average of 10 times faster results, and this test is 30 times faster on average. Times, up to 326 times the results, it can be said "unprecedented"!
The "computing power" of the diffusion analysis calculation engine
Today, SAS Viya has achieved the "king" status of the cloud-native analytical computing engine. The next task of SAS is to spread the capabilities of SAS Viya to more industries and fields. This is a solution based on SAS Viya support. Goodnight said that SAS not only has a powerful distributed high-performance analytical computing engine, but also solutions for various industries and scenarios. Solutions will be SAS's next investment focus and strong growth point.
At the 2023 SAS Innovate conference, SAS announced that it will invest $1 billion in industry AI solutions over the next three years. These investments will target banking, insurance, government, healthcare, retail, energy and other industries, using AI, machine learning and advanced analytics to solve challenges in these industries. In the past 40 years, SAS has rich experience in solution development, and has developed many star solutions with sales exceeding 100 million US dollars. These solutions have solved many common problems faced by many industries to a wider extent. , such as fraud predictive analytics that are required across multiple industries.
Bryan Harris, executive vice president and CTO of SAS, said in an interview that SAS's R&D investment mainly includes two parts: vertical integration and horizontal integration. Vertical integration is to integrate industry solutions with the SAS Viya platform, so that industry solutions can obtain super productivity, performance and security, as well as strong portability on the SAS Viya cloud-native platform. The SAS Viya cloud-native platform is based on containers and Kubernetes. In addition to being deployed in several super public clouds, it also supports hybrid cloud architectures such as Red Hat Openshift and Nutanix hyper-converged architecture.
Horizontal integration refers to the same scenarios for different industries, such as marketing analysis, transaction fraud analysis, supply chain analysis, etc. that are required by all industries. When SAS develops solutions, it pays more attention to the repeatable IP under the solution to maximize economic benefits. Of course, for industries with a large enough market, SAS also provides targeted solutions. For example, SAS Health is a solution for the medical and health industry, and SAS Clinical Enrollment Simulation is specifically to solve the problem of patient recruitment in clinical research. , SAS Customer Intelligent 360 helps brands use consumer data, marketing technology and resources to achieve more sustainable and responsible marketing.
SAS's US$1 billion investment in the next three years mainly includes: on the one hand, transplanting existing SAS industry solutions to SAS Viya; on the other hand, combining cloud, big data and AI to develop new capabilities, such as financial industry For enterprise decision-making, SAS connects the original isolated credit risk analysis, fraud analysis, marketing and other analysis systems to form a unified AI-based architecture and analyze on the basis of shared data, thereby helping financial institutions optimize operations across business chains , so as to achieve more accurate decision-making, improve customer experience and profitability.
Join forces with top partners
Committed to enterprise AI productivity, invest 1 billion US dollars in the field of AI solutions in the next three years, and partners are indispensable. SAS's AI solutions require powerful analytical capabilities such as SAS-based AI, advanced analytics, and predictive analytics, and most of these solutions rely on SAS's own analysis and data scientists and development teams, while also seeking to cooperate with top A strong alliance of partners.
Cooperation with Microsoft is a benchmark. In June 2020, at the SAS Global Forum 2020, SAS and Microsoft jointly announced that the two parties will form an in-depth strategic partnership to jointly shape the future of data analysis and artificial intelligence. Three years later, SAS technology has been deeply integrated with Microsoft Azure, Dynamics 365, Microsoft 365, etc. At SAS Innovate 2023, SAS once again announced the more in-depth results of its cooperation with Microsoft: SAS, Microsoft and AMD have cooperated to provide customers with SAS Viya services on the Microsoft Azure Confidential Computing privacy computing platform based on AMD EPYC chips, which can Meet the sensitive data security and privacy protection of enterprise customers in memory computing.
Another example of a technology partner is SingleStore. Bryan Harris grandly introduced the cooperation with SingleStore at the SAS Innovate 2023 conference, and set this cooperation as another partner benchmark. SingleStore provides an ultra-fast cloud-native HTAP database, which integrates data processing and analysis functions into a powerful distributed, multi-model database, and provides a unified database platform. The predecessor of SingleStore is the famous MemSQL, and it is also one of the proposers and advocates of the HTAP hybrid database concept. The biggest feature of SingleStore is the real-time database.
SAS and SingleStore began to establish a strategic partnership in December 2020, and in August 2022 launched the technical results of the deep integration of the two parties: SAS analysis technology can be directly used to analyze the data in the SingleStore database, due to the powerful SingleStore Distributed capabilities, user data can be analyzed on-site without leaving the storage location. Bryan highly praised the technical achievements of the deep integration of the two parties, and believes that this will be the next generation of analysis and AI architecture.
Building trustworthy AI is one of SAS Viya's visions. SAS has done a lot of work on Resposible AI, including: providing AI bias detection, explainability, decision auditing, model supervision, governance and trustworthiness on the SAS Viya platform; SAS Data Ethics Practice Team (Data Ethics Practice) has completed more than 200 consultations to help SAS companies and customers better practice trustworthy AI; promote trustworthy innovation training within the company; hold the US National Artificial Intelligence Advisory Committee meeting at the company headquarters, SAS Reggie Townsend, vice president of the company's data ethics practice, is a member of the committee.
The latest achievement of SAS Trustworthy AI is the establishment of the first Responsible and Ethical AI in Healthcare Lab (Responsible and Ethical AI in Healthcare Lab, REAHL) with two Dutch partners. The two Dutch institutions are European Erasmus University Medical Center, a top academic hospital, and Delft University of Technology, which has a Center for Digital Ethics. REAHL aims to create an AI ethical framework for medical and health institutions, which is applicable to medical institutions and health centers around the world. REAHL has also been upgraded to an official laboratory of the Dutch National AI Innovation Center.
SAS is also one of the founding members of the Coalition for Health AI (CHAI), which recently released a blueprint for trustworthy AI in healthcare. SAS also cooperates with Duke Health, the world's top academic medical center, to explore innovative data and analysis solutions for the healthcare field.
Towards an IPO: The Road Ahead
SAS plans to prepare for an IPO by the end of 2024. In the latest 2022-2023 annual report of SAS, SAS disclosed an annual revenue of more than US$3 billion, of which SAS Cloud sales increased by 14%, SAS Sales of Viya have increased by 23%, and 90% of Fortune 100 companies are SAS customers.
Gavin Day, senior vice president of the CEO Office in charge of SAS's IPO, revealed that SAS has formulated a three-year strategic plan and is steadily advancing its implementation. SAS' plan for "Beyond Tomorrow" is simple: support all languages, tools, and environments customers are familiar with, including R, Python, and SAS's own languages and low-code/no-code programming. For enterprise customers, SAS provides a unified analysis platform that manages all AI analysis, advanced analysis, and predictive analysis of the enterprise, and can run in any environment such as public cloud, private cloud, and hybrid cloud.
Gavin emphasized that partners are very important to SAS, and SAS has started to update the incentive policies for partners and distributors at the beginning of this year, so that partners can understand how to cooperate with SAS in a long-term and sustainable way, and have a deep understanding of SAS technology, Deeply cultivate SAS solutions to continuously drive business growth.
On the whole : SAS has completed the transition to the cloud-native era, and SAS Viya has once again become the "king" of cloud-native analysis. At the same time, SAS is still investing heavily in industry AI solutions, further driving the long-term sustainable growth of SAS. Just like a computer's operating system, SAS Viya is becoming the "operating system" for enterprise AI analytics. (Text/Ning Chuan)