The technological changes that occur in the field of generative AI are measured on a weekly basis.
In November of this year, the first phase of Amazon Cloud Technology Entrepreneurship Accelerator will conclude. From the initial recruitment to the current 3-month practice, Amazon Cloud Technology has collaborated with 28 venture capital and industry institutions, providing 6 modules and 28 courses, inviting more than 40 experts to focus on supporting 30 shortlisted startups in the AI field, helping them identify business scenarios in multiple tracks such as generative AI, enterprise service software, e-commerce solutions, and intelligent automotive solutions.
At the end of June this year, when the preparation for the entrepreneurship accelerator was just beginning, Gu Fan, the General Manager of the Strategic Business Development Department in Greater China, told a reporter from the Daily Economic News that being a startup accelerator and supporting startups in the field of generative AI could bring us some things that cannot be seen behind paper numbers in business growth.
After this practice, compared to these startups, Amazon Cloud Technology may have gained the most. The feedback from these entrepreneurial pioneers has given Amazon Cloud Technology the latest "feeling" of the implementation of generative AI in the entrepreneurial field and market. For example, the "Hundred Model Battle" is converging, and companies are more eager for large models to participate in specific businesses,
During an interview with reporters, Zhang Yutong from Jinshajiang Venture Capital stated that even though there is still a certain distance from the ceiling to the specific landing, there are constantly new opportunities emerging in the AI track and theme, which are areas that can be invested in for a long time. She believes that this does not affect the investment cycle.
After receiving these feedback, Amazon Cloud Technology has taken new actions and adjustments. Hu Ying, Director of Startup Ecology and Investment Business Development at Amazon Cloud Technology in Greater China, told the Daily Economic News that the second phase of the entrepreneurship accelerator is on its way, and the theme is still focused on generative AI. Among the enterprises targeted, in addition to those already served by Amazon Cloud Technology and those that have gone global with generative AI, there are also digital enterprises that have gone global.
One detail with four variations
In Gu Fan's view, one of the main things that this entrepreneurship accelerator does is "connect": the connection between generative AI startups and Amazon Cloud Technology, the connection between large enterprises and venture capital and industrial institutions, and the connection between startups. As a connecting platform hub, Amazon Cloud Technology has gained first-hand information on the needs of startups, B2B customers, investment institutions, and technology needs.
Gu Fan told a reporter from the Daily Economic News that there is a change: he previously talked about areas that are more technologically oriented, such as AI, big data, and cloud computing, and dealt with the CIO (Chief Information Officer) of the enterprise; But nowadays, when going out to meet clients, those who are more interested in generative AI technology are often the leaders of various business lines within the enterprise.
"Enterprises' understanding of generative AI has taken a step further than six months ago. Business line leaders will feel that this thing is visible and tangible for the first time," Gu Fan said.
The continuous expansion and clarity of cognition have made enterprises disenchanted with big models. More and more enterprises understand that big models are just "technology and capabilities", and begin to hope to make good use of this new tool to bring efficiency improvements at the business level. This change brings a result: the direction of the "Hundred Model Battle" is gradually converging.
From the changes in generative AI startups, Gu Fan summarized three points. Firstly, more and more startups are shifting their focus from basic models to application and toolchains. 95% of the start-up enterprises engaged in generative AI business are engaged in entrepreneurship related to generative AI applications and toolchains.
Secondly, generative AI products have highly digitized characteristics, such as accessibility, scalability, and are not limited to physical infrastructure and traditional channels, making them highly suitable for global development. Therefore, most startups in the field of generative AI are born global. In the acceleration camp, 96% of startups stated that they have plans to go abroad or have overseas layouts.
Thirdly, the products of startups face significant challenges in entering the market. Entrepreneurship in the field of generative AI is forming some ecosystems, and some companies are starting to form clusters and develop together with division of labor. At present, companies in Amazon's entrepreneurial accelerator are leveraging and collaborating with each other. Gu Fan said that the entrepreneurship of generative AI will take the form of ecology, and regular growth will also be one of the future trends.
What kind of results should the first phase of entrepreneurship accelerator achieve? Hu Ying, Director of Startup Ecology and Investment Business Development at Amazon Cloud Technology in Greater China, stated in an interview with the Daily Economic News that Amazon Cloud Technology measures the performance of accelerators from three dimensions internally: are the participating companies satisfied? Has co creation achieved its goals? Is the specific project feasible at the commercial level?
At present, this entrepreneurial accelerator is generally satisfactory, and the second phase of the accelerator is already planned.
From Imagination to Implementation: Entrepreneurial Opportunities in Taming New Tools
Zhang Yutong, the managing partner of Jinshajiang Venture Capital, stated in a sharing that the explosion of generative AI has brought two exciting points to Jinshajiang Venture Capital. Firstly, the ability to generalize AI, which means that economies of scale can be formed in the future; The second is that super AI is a super UI (user interface), and the new interaction methods mainly based on "language, dialogue boxes, and prompt words" can bring a lot of new experiences.
Indeed, the technological changes that occur in the field of generative AI are measured on a weekly basis; But when it comes to specific business scenarios, the time required is measured in months or even years.
Gu Fan said in an interview with the Daily Economic News that the large model technology determines the ceiling of the industry's growth imagination space, but at the specific implementation stage, the gap between imagination and reality is "unimaginable.". From technology to specific applications, it's like dreaming from imagination to building bricks and tiles, which requires a lot of time. The road to truly solving generative AI applications in business is still very long.
One fact is that currently no enterprise can fully implement customer generated AI applications with just one solution and product.
"This is like a 'hamburger', not three layers, it could be five or six layers. However, at the same time, many layers of this' hamburger 'are not interconnected. If these empty spaces are not filled in, it will be difficult for generative AI applications to be universal," said Gu Fan. He stated that currently, data engineering, model tuning, model inference, and other aspects are the challenges for inclusive use of generative AI applications in enterprises.
"Today, a wave of startups will provide their core values based on the gaps in this' hamburger '. Therefore, we also hope that startups can establish connections, form complementary advantages, and jointly develop solutions that meet the different needs of customers." Gu Fan said.
The huge gap between imagination and implementation is also a huge space for startups to create value. This represents development opportunities and investment returns. Jinshajiang Venture Capital's Zhang Yutong told reporters that the industry has seen good results so far.
"For example, ChatGPT has 100 million weekly active users; Midjournal has a team of only 11 people and has not raised funds, but its revenue in the first year reached 100 million US dollars, and it is expected to reach 200 million US dollars this year. Whether in the field of cultural graphics or large models, we have seen a user group with sufficient stickiness and early commercialization results. These commercialization results are also constantly penetrating different fields," said Zhang Yutong.
"The development of generative AI is like a marathon race, and the race has just begun. For startups, this is also a brand new era," said Gu Fan.