How ByteDance Should Respond to OpenAI’s Sora – and Text-to-Video GenAI (2 of 2) ( Tech Strategy – Daily Article)

Posted on by

This is Part 2 for the 4 steps I would focus on ByteDance, with regards to the rapid rise of text-to-video generative AI. Part 1 is here.

Step 2: ByteDance Should Use Its Superior Scale and Resources to Win Long-Term

ByteDance is behind in generative AI. But that’s not terrible if you are a large tech company with tremendous resources, technology, and cash flow. They can do something that most companies cannot do. They can basically make large investments and sustain them for a long-term. Generative AI is ultimately a marathon, not a sprint.

ByteDance has +150,000 employees in 30 countries. And revenue for 2023 is reported to be $110B (with pretty great margins probably). If they invest 10% into R&D for generative AI, that’s $10B per year. And over five years, that’s $50B. Yes, Baidu is arguably the generative AI leader for China today. But their annual revenue is closer to $20B.

Note that Huawei increased their R&D spending to 25% during their crisis. And they plan to maintain that for 10 years.

ByteDance has the ability to do big and long-term investments, which can be in hiring, R&D, and partnerships. And maybe some M&A. I would be pulling all of these levers.

Step 2: ByteDance Should Use Its Superior Scale and Resources to Win Long-Term

And doesn’t this also sound like CEO Liang Rubo’s first and third objectives for GenAI – to “strengthen the recruitment of AI talent” and “improve fundamental research”.

However, spending big on a new technology doesn’t necessarily get you results. It depends on having an innovative and entrepreneurial culture. So, I would do #1 (re-invigorate the leadership and culture) before #2 (spend big).

Step 3: Get GenAI into All the Core Products ASAP.

Baidu had their company meeting in October. And Alibaba Cloud had its Big Data and AI Summit a few weeks ago. Both presented what they are doing in generative AI. And both said the same thing. They are both putting generative AI into all of their company’s products ASAP.

These are really two of the companies to watch for GenAI in China. They both have been building their own AI tech stacks for a long time – with apps, training tools, foundation models, data products and even semiconductors. And they are both in the AI cloud business, so they really have to create an entire suite of “intelligence” services for their clients.

At Alibaba Cloud, they are Qianwen (Qwen), their suite of open-source foundation models, into pretty much every product at Alibaba. From Taobao and Youku to Ele.me and Cainiao.

At Baidu, they have prioritized their AI Cloud business. And they are also integrating their suite of open-source foundation models into their products and services. Baidu has a smaller range of products than Alibaba, but they have a much larger suite of foundation models and applications today. And they have a really smart strategy focused on creating industry-focused foundation models that create a knowledge flywheel between industry usage and model customization. It’s worth looking at what they are doing. You can hear my breakdown here.

But the key step CEO Robin Li has been talking about is rebuilding Baidu’s family of apps (which they call their Mobile Ecosystem) with GenAI. This includes:

  • Baidu Search. Their flagship app for search, feed, content, and other services through mobile devices. It has “twin-engine search and feed functions that leverage our AI-powered algorithms and deep user insight to offer users a compelling experience…In December 2020, MAUs and DAUs of Baidu App reached 544 million and 202 million, respectively.”
  • Baidu Maps. This is listed under Baidu’s Intelligent Driving business. It’s “a voice-enabled mobile app providing users with travel-related services, including POI search, route planning, precise navigation, taxi-hailing service and real-time traffic condition information. Baidu Maps has a MAU of 316 million in December 2020.”
  • Baidu Wenku
  • InfoFlow
  • Baidu Drive
  • Baidu-owned iQIYI is also very aggressively integrating generative AI into its core content production. More on that in another article.

There were reports in late 2023 that ByteDance did launch its chatbots Doubao and Cici AI. But that is pretty standard stuff at this point.

I actually think ByteDance should copy Microsoft and do GenAI in their B2B apps. Lark and Volcano Engine are interesting products that are going international. Plus, Volcano Engine is just a great name for a product.

Lark Suite is their enterprise collaboration and communication tool suite. Similar to DingTalk and Microsoft Teams. It was released in 2019 and took off during Covid when everyone started working from home (and doing video conferences).

Note: Most of the China tech companies launched similar services during this period (WeChat Work, Alibaba’s DingTalk). Lark has increasingly been going international.

Volcano Engine was another clever externalization and then internationalization move. As mentioned, ByteDance is good at rapidly creating, testing, and deploying mobile apps. They turned the internal tools they use to do this into a service, which they call Volcano Engine. Basically, it is a cloud service with lots of strengths in algorithms and data science. And it can be used in a fairly lightweight manner, which lets you ramp up and test apps easily. The international version is called Byte Plus.

Both of these apps are good candidates for generative AI. I think they can pretty much copy what Microsoft is doing with OpenAI.

But the big focus is obviously their core short video app.

Step 3: Get GenAI into All the Core Products ASAP.

***

Step 4: Get GenAI into Creator Tools and Services ASAP.

I really like CapCut. It’s ByteDance’s “all in one” video editing app that is a huge hit in the US and internationally. I used to use Adobe Premiere on my PC for videos. Now I just use CapCut on my phone. It makes editing videos really easy (which you can then upload to TikTok).

And they keep adding features:

  • You can add text.
  • You can add music and sound effects.
  • You use their growing library of filters (free and paid). There is the filter that can de-age you (which I avoid). There is the filter that can put a wig and make-up on you (which people are using to troll Bud Light). And there is the video that makes you look like a model.

CapCut is a good example of how they provide value to content creators. Like few other companies do. ByteDance has a growing list of tools and activities for content creators. They even host events and create activities for them. This includes trips to scenic locations (that are good for making videos).

These tools and activities are particularly effective when they are entering new markets. When entering the US, TikTok was very good at supporting local content creators. Especially those that had been treated badly by Facebook, Instagram, and YouTube over the years.

In my mind, this is what makes ByteDance difference than Alibaba, Baidu and the other companies mentioned. They are ultimately a creative company. They are enabling creativity in their users. Their front lobby talks about creativity and inspiration.

And creativity is what generative AI is best at. It’s not that good at getting the right answer to questions. But it’s very good at creating poems, images, videos, and other things where there is no definitive right answer.

Generative AI is a creator tool. It is the best one that has ever been created. That is why it is such a big potential disruption to ByteDance’s core business.

It’s also why ByteDance is so well positioned to do well in generative AI. This is ultimately going to come down to who is better at empowering the creators of short videos and other content. And ByteDance is really good at creating and integrating new creator tools. In this sense, this is just one more. But it’s the big one.

SCMP has reported that ByteDance is building an in-house video motion control tool called Boximator. It is designed to help content creators make videos. That’s the core GenAI Capability they need to build. But SCMP says it’s still early stage.

GenAI enabled tools for content creators is the #1 thing I am watching for at ByteDance this year. And that’s Step 4.

Step 4: Get GenAI into Creator Tools and Services ASAP.

***

Ok, that’s it for this article. It’s an exciting topic and I’m pretty optimistic.

Cheers, Jeff

————–

Related articles:

From the Concept Library, concepts for this article are:

  • Generative AI: Text to Video

From the Company Library, companies for this article are:

  • ByteDance
  • OpenAI: Sora

Photo by Alexander Shatov on Unsplash

———-

I write, speak and consult about how to win (and not lose) in digital strategy and transformation.

I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.

My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.

This content (articles, podcasts, website info) is not investment, legal or tax advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. This is not investment advice. Investing is risky. Do your own research.

Leave a Reply