You have clicked a link to access information on a new website, so you will be leaving Because this new site is independent, Diamond Hill Capital Management, Inc. neither manages nor assumes responsibility for its content. Are you ready to move forward?

How Machine Learning Is Creating Value for Google and Facebook

By Varun Gupta, CFA
June 2020

Over the past several decades, computers have become very fast, but typically not very smart—they haven’t had much ability to learn from data, and instead needed precise instructions to carry out a task. More recently, machine learning and artificial intelligence (AI) techniques have leveraged computer algorithms to attempt to mimic human brain processing. When given vast amounts of data, these algorithms can learn a task, producing superior insights over time as the algorithms improve. Today, machine learning and AI technologies are everywhere—displaying better search results on Google, prioritizing content we like on Facebook, filtering out low-quality and spam content on the internet and improving digital advertising efficiencies for businesses. These technologies are also leveraged outside of the tech space, solving problems for companies across industries, such as retail and health care.

Interview with the Author


The views expressed are those of Diamond Hill as of June 2020 and are subject to change. These opinions are not intended to be a forecast of future events, a guarantee of results, or investment advice.

As of May 31, 2020, Diamond Hill owned Facebook, Inc. – Class A (equity), Alphabet, Inc. – Class A (equity), Tencent Holdings Ltd. (equity), Berkshire Hathaway, Inc. – Class B (equity), Walt Disney Co. (The) (equity) and Microsoft Corp. (equity). As of March 31, 2020 Diamond Hill owned Netflix, Inc. (debt), Walt Disney Co. (The) (debt), Berkshire Hathaway Financial (debt) and Apple, Inc. (debt).

In our view, there are three key technological forces that have converged to accelerate the growth of machine learning and AI over the past few years.

  1. The plummeting cost of computing and computer storage has made execution of computationally intensive and data-hungry algorithms cost-effective.
  2. The rise of cloud computing and penetration of high-speed internet have provided access to high-performance hardware and software infrastructures needed to run highly computational and storage-intensive machine learning algorithms. Cloud computing allows companies to access, or rent, the sophisticated technology needed without deploying significant capital to develop their own infrastructure.
  3. The explosive growth of digital services on intelligent devices, such as smartphones and voice assistants, has resulted in enormous data volumes being generated and captured. These large data sets provide machine learning algorithms with abundant resources to learn and find patterns, improving the algorithms over time.

By leveraging these three technological trends, machine learning algorithms can build more precise models— helping companies affordably identify profitable opportunities, avoid risks, gain insights and enable new applications.

Machine learning and AI algorithms use various techniques to find patterns in data. Anything that can be digitally stored—such as text, numbers, images, clicks, audio and video—can be fed into machine learning algorithms. These algorithms are automated and improve with minimal human intervention when exposed to massive amounts of data. Apps and services from Alphabet and Facebook have attracted large user bases over time—as users interact with these apps and services, they generate real-time data which help machine learning algorithms extract insights, driving continual product improvements and the development of new features. This constant product improvement based on user data makes it hard for competitors without access to real-time data to catch up. While machine learning and AI raise privacy and societal concerns among some users and regulators, both Alphabet and Facebook have taken actions in recent years to demonstrate they are at the forefront of using these advanced technologies responsibly and ethically, working collaboratively with different stakeholders to combat media sensationalism that can be associated with new technology trends.

We believe the use of machine learning enables a positive feedback loop, or virtuous cycle, for users, advertisers and businesses, including Alphabet and Facebook. Users benefit from high quality content and the removal of spam and irrelevant ads. Advertisers benefit by simplifying and automating complex functionalities, including:

  • Locating and reaching the right audience within Alphabet and Facebook’s vast user base.
  • Creating custom ads using text, images, videos or a combination of mediums, generating multiple versions to test which combination performs best.
  • Placing ads effectively across a wide range of Alphabet web properties (Search, YouTube, Maps and Gmail) and Facebook apps (Facebook, Instagram, Messenger and WhatsApp).
  • Automating the ad-bidding process to ensure realtime placement.
  • Receiving data-driven insights to ensure they are spending their budgets effectively and unearthing and capitalizing on new opportunities.

These functionalities allow advertisers of any size to benefit from technologies that may not otherwise be available to them, increasing their return on investment. Contrary to concerns that machine learning algorithms are replacing the need for human advertisers, we regard the trend as a positive development and view these technologies as tools for advertisers rather than a replacement. Machine learning algorithms perform tedious and mundane tasks, freeing up human advertisers for strategic thinking. As advertisers obtain more value from Alphabet and Facebook’s advertising offerings, they tend to bid higher in the ad auction, enabling both companies to fairly extract better pricing for the additional value they have created through machine learning techniques. These techniques also allow Google and Facebook to provide their advertising services globally and scale their advertising features across their diverse customer base, including new advertisers.

As long-term, intrinsic value-based investors, we constantly monitor how a company’s competitive moat might change in the future. History shows that moats for technology companies can be in jeopardy when new innovations driven by technological advances are brought to market by competitors. We continuously monitor technological changes and carefully evaluate how they influence our estimates of intrinsic value. We expect Alphabet and Facebook to continue to lead in machine learning and AI techniques, leveraging the technology to provide superior user experiences and best-in-class, self-service advertising platforms for advertisers.

As of 5/31/20, Diamond Hill owned Alphabet, Inc. (Cl A) (equity) and Facebook, Inc. (Cl A) (equity).

Originally published on June 24, 2020.

The views expressed are those of the research analyst as of June 2020, are subject to change, and may differ from the views of other research analysts, portfolio managers or the firm as a whole. These opinions are not intended to be a forecast of future events, a guarantee of future results, or investment advice.

back to top