The 4 Pillars of Baidu’s AI Cloud Strategy (3 of 4) (Tech Strategy – Daily Article)

Posted on December 28, 2023 by Prof Jeff

Baidu AI Cloud revenue reached 8.4B RMB in Q4, with generative AI and foundation models accounting for 656M RMB.

That was a surprise.

First, they did well.

And second because we haven’t really seen any GenAI-driven revenue numbers from other companies. Baidu also gave some interesting numbers on GenAI operating performance.

ERNIE (their suite of foundation models) had + 26,000 monthly enterprise users on cloud.
ERNIE was getting 34,000 queries per minute in December.
The inference cost of ERNIE 3.5 has dropped to 1% of what it had cost in March 2023.
Qingge (Baidu’s AI native advertising platform) has reached +10,000 advertisers.
Apollo Go (their autonomous driving platform) delivered 5M orders, a 49% increase in Q4. And driverless rides has reached 49% of orders in Wuhan (one of the test cities).

It looks like they are really moving in their AI Cloud business.

But let’s get to the strategy breakdown (my area). I think Baidu’s AI Cloud strategy has 4 main pillars.

Pillar 1: Baidu AI Cloud Is an Innovation Platform Pushing to Reach Critical Mass
Pillar 2: Baidu AI Cloud Is Also Moving to Become a Technology Standard Deeper in the AI Tech Stack
Pillar 3: Baidu’s AI-Focused Cloud Has 2 Advantages Over Other Innovation Platforms
Pillar 4: Baidu’s AI Cloud Is a Flywheel in Industry-Specific Intelligence

Let’s start with number 1.

Pillar 1: Baidu AI Cloud Is an Innovation Platform Pushing to Reach Critical Mass

That’s the biggest thing happening here.

Baidu AI Cloud is building an innovation platform.
And on top of that they are planning on keeping a few of the key apps (i.e., the crown jewels). These will either be strategically important or cash cows.

Microsoft did the same thing with Windows. They have the operating system, which is an innovation platform and a monopoly. They also kept Microsoft Office, which is a cash cow. And has become a strategic capability in enterprise.

Apple did the same thing with iOS. They own the operating system (an innovation platform). And they kept Apple Maps, which is key strategic capability.

That is at the center of what Baidu is doing in AI Cloud. You can see in their mobile ecosystem what they are likely going to keep control of on the platform. Definitely search. Also, the driving operating system DuerOS. And definitely mapping. We’ll see what else.

Pillar 2: Baidu AI Cloud Is Also Moving to Become a Technology Standard Deeper in the AI Tech Stack

Innovation platforms, like Microsoft Windows and Apple iOS, sit on top of the tech stack. And Baidu cloud is doing this. But they are also moving deeper into the AI tech stack. They are building the capabilities all the way down to the semiconductors.

So, they are also doing a standardization and ecosystem strategy. This makes them closer to ARM Holdings. They are creating an architecture for all of AI. And you can see Baidu pushing to try and become the technology standard all the way down.

To do this, they need to have lots of integrations with other technologies in the tech stack. They need hardware and software makers to all integrate with their foundation models and semiconductor architecture.

They also need lots of applications built on their tech stack. As part of the innovation platform. But also for deeper in the tech stack. So they they need to encourage and train lots of developers and companies. And not just at the application layer.

You can see them doing this by:

Making their technology open source. They want everyone to build on it. And integrate with it.
Encourage developers and companies to build on their tech stack.
Putting their AI tools and technology into all of their Mobile Ecosystem apps. This is to jumpstart adoption.

The key is to get usage – primarily by developers, business partners, and customers. As mentioned, ERNIE (their suite of foundation models) has + 26,000 monthly enterprise users on cloud. Baidu also cited some top tier companies as their customers / partners, including Samsung China, Autohome and Honor.

They announced a strategic partnership with Samsung Electronics China. This will debut with the launch of the Samsung Galaxy S24 series, “a flagship line equipped with rich AI applications, powered by Baidu’s ERNIE Foundation Model.”

Integrating with smart device manufacturers is a big growth path. Baidu says the “integration includes a variety of on-device AI functionalities such as Al Live Translate for call translation, as well as generative AI experiences like intelligent summarization and advanced typesetting, equipping the Samsung Galaxy S24 series with more robust AI application scenarios.”

Watch for lots of other strategic partnerships.

And finally, for this strategy to work, they need to grow AI talent. This is probably the biggest limitation on growth rate. So, they have their 5M AI Talents Program.

Baidu Is Putting their AI Tech into All Their Products ASAP

As mentioned, Baidu is trying to get as many developers, business partners and customers to use and incorporate their AI tools and technology as possible.

And you can jumpstart this by putting your tech in your own products and services. Which they are doing very aggressively.

At the Baidu event in October, CEO Robin Li showed this. He went through a list of all the applications internally and externally that ERNIE is being incorporated in – including:

Baidu Maps. They claim this is now the first AI-native mapping app.
Baidu Drive. This now has a personal AI assistant built in.
Baidu GBI (Generative Business Intelligence). They are using generative AI for simplified interactions for data queries and analysis.
Baidu Wenku. This is being transformed from a content documenting platform to a “content productivity platform”. Baidu claims it can now “assist users in conducting academic research, summarizing key information, creating documents, and generating slideshow presentations, drastically reducing the time spent in content production in scenarios such as preparing a keynote speech.”
Infoflow Super Assistant. This is an AI assistant that streamlines communication, collaborations, knowledge management and business services.

Those are cool. But Baidu’s big guns are in search and marketing. Those are the products to watch.

Baidu Search Is Incorporating AI. Important.

Baidu says its search (with AI) can now do “enhanced recommendations”, “multi-round interactions”, and “precision excellence”.

That’s pretty interesting.

They say search queries can now generate multimodal answers with text, images, and charts. So, links as well as generative content as answers to search queries.

Multi-round interactions means the search engine can continue the interaction from your current queries. That is an improvement. But it is also an interesting type of personalization. It is able to respond to multiple prompts and make adjustments based on what you really want. It’s highly detailed type of personalization.

I’m not really sure what they mean by “precision excellence”.

Baidu Marketing Is Incorporating AI. Also Important.

There are at least three interesting things happening here.

Brand BOT – which is a new conversational brand marketing product. This lets companies create their own digital avatars for their brand These avatars can do multi-round interactions with customers.

In addition to enabling “a brand-specific model” for companies, it also enables “brand personality capabilities”. Giving companies and products different personalities is really interesting.

Qingge – which is basically an AI-native marketing agency within the Baidu ecosystem. It’s an AI service that can do insight, analysis, and production in marketing for companies. Baidu says that with Qingge the “average ad creation time has been shortened from 2 hours to 5 minutes, the average conversion volume has increased by 10%, and the number of interactions has reached one million.”

And if you integrate Qingduo, customers can do copywriting, image generation, and digital human video production at the same time.

Qingduo – which does multimodal content generation (text creation, image creation, and video creation). Like discussed for iQIYI, AI-enabled content generation can both increase quality (better and more effective content) and improve efficiency (i.e., cheaper and faster). Baidu says Qingduo can “generate 100 creative copies in 2 minutes, create a digital human model in 3 minutes, and produce a complete digital human spokesperson video in 5 minutes.”

***

Ok. Moving on to the next pillar of their AI Cloud strategy.

Pillar 3: Baidu’s AI-Focused Cloud Has 2 Advantages Over Other Innovation Platforms

So, it’s an innovation platform. Plus, a play for standardization in other parts of the tech stack. That’s a powerful business model. But there are others trying to do the same thing.

How do you compete with them? How can Baidu’s AI Cloud compete with giants like AWS, Google and Alibaba Cloud?

There are definitely two clear advantages at this point.

Baidu’s first advantage is they are an early mover in their AI tech stack and innovation platform.

Baidu was an early mover. And they are getting adoption. Innovation platforms and tech standards are a lot about who gets to critical mass first. It’s a game of demand side scale. And we can see that Baidu is fairly far down the path on this already. There is a growing list of apps and use cases.

Baidu’s second advantage is their self-developed native AI tech stack.

Baidu is not adapting an existing digital tech stack for AI use cases and capabilities. They are not repurposing chips, server stacks or software operating systems. They have created entirely new technology that is built specifically for AI from day one. And this includes both the software and hardware, which are integrated at every level. And which are vertically integrated across the stack.

Baidu AI Cloud is AI-native. This should make their services more efficient and effective (cheaper and better, in theory). And it should also give them a real advantage in development. They should advance more rapidly and more efficiently than others. Baidu management cites this – more efficient development from the integration of Cloud and AI – as one of their big advantages.

From Baidu:

“The integration of Cloud and AI is a unique advantage of Baidu AI Cloud. As a cloud tuned for AI, it provides a safe, stable and flexible infrastructure for digital transformation, while the “AI engine” provides leading innovative technologies and platform to drive intelligent upgrading…It is the seamless integration of the “cloud tuned for AI” and the “AI built for common scenarios” that build the cornerstone of modern, intelligent transformation.”

***

And that brings me to the fourth and final pillar of Baidu’s AI Cloud strategy. They are aggressively building a flywheel in industry-specific intelligence.

Pillar 1: Baidu AI Cloud Is an Innovation Platform Pushing to Reach Critical Mass
Pillar 2: Baidu AI Cloud Is Also Moving to Become a Technology Standard Deeper in the AI Tech Stack
Pillar 3: Baidu’s AI-Focused Cloud Has 2 Advantages Over Other Innovation Platforms
Pillar 4: Baidu’s AI Cloud Is a Flywheel in Industry-Specific Intelligence

I’ll cover that in Article 4.

Cheers, Jeff

——-

Jeffrey Towson 陶迅

Consultant and keynote speaker on digital and AI strategy, transformation and disruption