OPPO Acquires LM Startup Company AIWaves

On October 22, Jiemian News learned from insiders at AIWaves that the company will be acquired by the mobile phone manufacturer OPPO, and CEO Jiang Yuchen will join OPPO. OPPO told Jiemian News: “Currently there is no more information available.”

“We are just being acquired, our products are still operating normally, it’s not like what has been circulating online.” The above-mentioned person told Jiemian News.

AIWaves was established in 2023, with a registered capital of 1.34 million yuan and a paid-up capital of 240,000 yuan. The actual controller is Hangzhou Wuliang Enterprise Management Partnership Enterprise (Limited Partnership), and the founder Jiang Yuchen holds a total shareholding ratio of 36.42% through platforms and individuals.

Public information shows that Jiang Yuchen was born in 1998, graduated from the Gong Gao Class of Zhukezhen College at Zhejiang University with an undergraduate degree, and obtained a doctoral degree from ETH Zurich under the supervision of Prof. Ryan Cotterell, specializing in natural language generation.

In January 2024, AIWaves completed a Pre round A financing of tens of millions of yuan, led by Lanchi Ventures, with West Lake Technology Innovation Investment, and Ant Financial Chairman Jing Xiandong as co-investors. At the same time, AIWaves released the self-developed Chinese creative domain large model ‘Weaver’ and the user-writing Agent product ‘WawaWriter 1.0’ driven by it.

The Weaver large model was jointly developed by AIWaves, APUS, and Professor Chen Huajun’s team from Zhejiang University.

Jiang Yuchen once stated that the Weaver large model is specifically designed for creative purposes, with its main feature being pre-training on 50B tokens of high-quality Chinese content, combined with continuous adjustments by engineers to make the writing content of this large model more ‘human-like’.

Jiang Yuchen mentioned that the team developed RecurrentGPT, a large language model capable of simulating RNN/LSTM (Recurrent Neural Networks/Long Short-Term Memory networks), enabling interactive generation of extremely long texts and making it possible for the large model to create lengthy novels.

At the end of July, AIWaves released a new generation personalized adaptive private language model Weaver 2.0 for generating multi-modal infinite-length content, along with upgraded multi-language multi-modal AI content creation tools such as WawaWriter 2.0, Siuuu.AI, AI Learning, and other series products.

‘From infinite text generation to infinite content generation, WawaWriter has created a multi-modal empowered video storytelling tool that effortlessly covers full-chain content creation from novels to scripts to videos encompassing text, audio, and visual elements,’ said Wan Lei CPO of AIWaves. The addition of LPA technology makes WawaWriter a ‘the more you write, the better it understands you’ personal assistant.

At that time, CPO Wan Lei referred to the above-mentioned upgrade as a ‘transformation and upgrade of the one-stop full-chain creation tool platform.’

However, less than three months after the release of its new model, AIWaves was acquired by OPPO, casting doubt on the commercialization model of long-text generation in this field.

The main profit models for WawaWriter are subscription memberships and one-time service billing. The product is positioned towards professional novel authors, self-media operators, and other content producers to improve writing efficiency. The willingness and ability of these users to pay for services, including market space, still need validation.

On the other hand, long-text generation based on large language models (LLMs) has huge potential in automated content creation and data processing, but still faces a series of technical challenges, as well as issues such as heavy consumption of computing resources during model training and inference processes.

Since the second half of this year, there have been fewer financing news from large model companies. Compared to general large model companies, vertical-specific large model companies have received less funding and have fewer resources. General large models are struggling to make breakthrough progress. Not only are the ‘Big Six’ facing varying degrees of scrutiny, but vertical-specific model companies that were expected to commercialize are also beginning to face survival issues.

SEE ALSO: OPPO Find X8 Will Be Released on October 21 at the Earliest

During a period of slowing technological innovation, startups lag behind big tech companies in terms of funding, technology, and talent density. Previously, Zhu Xiaohu stated in an interview with Jiemian News that the best outcome for the ‘Big Six’ would be selling out to big tech firms.

Talent is also one reason why big tech firms choose to acquire startups. In the acquisition by AIWaves mentioned above, Jiang Yuchen joining OPPO serves as evidence.

The wave of mergers and acquisitions involving large models has already begun in Silicon Valley in the United States. In March this year Microsoft recruited key talents including Mustafa Suleyman from AI startup Inflection. After Amazon acquired AI startup Adept, its founders Shazeer and DeFreitas returned to Google.

In August, artificial intelligence startup Character.AI was acquired by Google; its founders Noam Shazeer and Daniel De Freitas will return under Google’s DeepMind department along with over 30 employees from Character.AI.