Google's AI Expansion: A Concerning Trend?

Google has announced its next step in artificial intelligence (AI) with “AI mode,” a new feature for its search engine users in the United States. There is no set date for when it will be available to users outside the US. The company describes it as being able to have a conversation with a knowledgeable expert on many subjects. This is one of many steps Google is undertaking as part of its “all-in” approach to AI.

The “all-in” approach includes integrating the technology into different applications. Google is providing products along the AI supply chain through “vertical integration,” managing everything from AI computer chips to user interfaces like Google Maps and Gmail.

Google isn’t alone in its vertical integration ambitions. OpenAI recently acquired a hardware startup, centralising hardware development. Amazon is also taking steps, using its cloud computing platforms, custom chips and device plans while adding more AI services into products.

This could be the start of a vertical integration trend across big tech, impacting users and companies.

Understanding the AI Tech Stack

Hardware, software, data sources, databases, and servers make up the “AI tech stack.”

Google's Tech Stack Layers

Google’s vertical tech stack consists of four layers:

1. Hardware layer: Google develops its AI chips, called tensor processing units (TPUs), which it says provide better performance and efficiency than general-purpose processors.
2. Infrastructure layer: Google uses its cloud infrastructure for computing power, networking, and storage to run and scale AI.
3. Model development layer: Google uses in-house research for machine learning, robotics, language models, and computer vision.
4. Data layer: Google collects data from users across platforms like its search engine, Maps, and Gmail as a condition of using its applications.

Some argue vertical integration is a cost-effective business strategy for many industries. However, Google and OpenAI dominate the global technology market, which allows them to charge higher prices and misuse online advertising practices. Vertical integration centralises the AI tech stack to one company, further skewing this power imbalance.

Distributing hardware, infrastructure, research and development, and data across multiple industries supports a more equitable playing field, and its loss creates barriers to entry for smaller companies. It also reduces incentives to innovate in ways that benefit consumers because it eliminates business competition.

The Impact of Data and Exclusivity

Data is very important in AI. Through its platforms, Google has access to a continuous stream of data, giving the company more power. The success of vertically integrated companies relies on in-house knowledge and expertise, and keeping this within a few companies can lead to knowledge hoarding. Research indicates this reduces social learning and increases disparities between market “winners” and “losers,” creating a vulnerable industry because net gains are lost in the pursuit of exclusivity.

Exclusivity also creates a lack of resilience by centralising failure points. Risk management requires oversight, transparency, and accountability. Centralising the AI tech stack within one organisation reduces external scrutiny by limiting interactions with external providers. This can lead to risky behaviour.

Regulatory bodies can provide external scrutiny, but the push to deregulate AI is widening the gap between technology development and regulation, allowing big tech companies to become opaque. A lack of transparency raises concerns about data practices. The trend towards vertical integration in the AI sector will increase this opacity and heighten issues around transparency.