Microsoft 'Future of Work with AI' (part 2): Why Microsoft will be a future Platform of Platforms
Paid article available for free
Microsoft's The Future of Work With AI event showcased innovations that have major implications for the enterprise software industry as a whole.
For a recap of the features, see part 1 (link)
A Shift in Software Paradigms
To understand what implications this might have for enterprise software, we need to delve into history.
The creation of higher layers of abstraction has generally led to significant shifts in profit capture throughout the technology value chain:
Computer hardware: in the early days of computing, hardware manufacturers made substantial profits on their PCs. However, each PC had a unique operating system interacting with the physical components, making it challenging for developers to create applications for all PCs in the market.
Operating system: the development of operating systems, notably Windows, marked a significant paradigm shift. Operating systems abstracted away the complexities of interacting with the PC’s hardware layer. Windows was also able to be installed on every PC hardware, providing a common interface for developers could build on. As a side effect, this commoditized all PC hardware down to its specs and cost. With little differentiation, profit margins in PC hardware were squeezed, while Windows reaped the rewards.
Web browser: the advent of the World Wide Web and web browsers resulted in another paradigm shift. Instead of developing applications for the operating system, both consumer applications (e.g., YouTube, Google) and enterprise applications (e.g., Salesforce) started being developed for the cloud, enabling users to access them via a web browser. As web browsers could be installed on any operating system, the importance and lock-in of operating systems diminished, marking the end of Microsoft's dominance.
Higher layers of abstraction also extend beyond technology:
Electrical power grids serve as a higher layer of abstraction, as the complexities of power generation and distribution are abstracted away from the user. In the early days of electricity usage, users needed to generate their own power through local generators or batteries. Electrical power grids allowed users to access electricity through a standardized interface, the wall outlet, without worrying about the complexities of power generation, transmission, and distribution.
Online travel agencies (OTAs) represent a higher layer of abstraction in the airline industry. Before OTAs, users had to engage with airlines directly, giving airlines greater control of customer relationships. After the advent of OTAs, airlines ended up mainly competing on cost. A large share of the profits also shifted away from airlines to the OTAs.
Embracing Chat Interfaces: The Next Paradigm Shift in Enterprise Software
Before applications, organizations used spreadsheets to do their work.
Applications introduced a graphical user interface (GUI) as an abstraction layer above spreadsheets (i.e. an application is just a graphical user interface sitting on top of a database). Users operated at the GUI level to facilitate data manipulation (read and write data) and made the underlying data structures invisible to users.
Large language models like ChatGPT have the potential to serve as the next abstraction layer above enterprise software applications. Instead of interacting with a GUI, workers will increasingly accomplish tasks and access information through chat interfaces, e.g.
"What was sales in X region for March?"
“Prepare me for my next meeting”
"Set up a meeting with X and Y and find the earliest availability”
To serve as a higher level of abstraction, it must reduce complexity for the end-user and ultimately be easier to use. This is likely the case for AI chat interfaces:
Chat interface has almost zero learning curve as it uses something that we already know - language!
You can use chat to orchestrate actions across multiple applications (as opposed to constantly ‘alt+tabbing’ across applications)
A few words can represent many actions, e.g. “add slide animations to this PowerPoint”
You can perform actions that can only be expressed in words, e.g. “make the wording of this email more formal”
The Commoditization of Applications
As the complexities of an application’s GUI gets abstracted away by a chat interface, the importance of GUIs as a differentiating factor will fade.
Applications will simply be left to compete on functionality and cost only
For example, Salesforce's user interface and ease of use differentiate it from legacy incumbents like SAP and Oracle. However, if the GUI is abstracted away from the end user and CRM interactions occur only through an AI chat interface (e.g. Microsoft Viva Sales), then this levels the playing field
This commoditization of the application layer will shift the plane of competition to the AI layer, e.g. organizations may start choosing applications that work better with the AI system (e.g. Microsoft 365 Copilot) that they’re using.
The Nash Equilibrium for Best-of-Breed Solutions
Best-of-breed or point solution providers must choose between integrating with broader AI systems or risking obsolescence.
Initially, best-of-breed solutions may try and create their own AI functionality.
However, organizations are more likely to choose applications that already work with their existing AI system, e.g.
A user of Microsoft 365 Copilot will likely choose an application that integrates well with it, rather than one which doesn’t
Consequently, most best-of-breed solutions will ultimately be forced integrate with all major AI systems.
This is a major shift in bargaining power from the application layer to the AI layer which will likely have profit share implications.
The End Game
The end game could be a single AI system that orchestrates work across all applications, i.e. Platform of Platforms! However, it may take many years before this end state plays out.
Until then, we may see smaller, specialized ‘platform of platforms’ emerging, e.g.
Front office (Salesforce, Hubspot)
Back office (SAP/Oracle, Intuit)
IT (ServiceNow)
Cybersecurity (Crowdstrike, Palo Alto)
Data (Snowflake)
Over time, the market could consolidate to just a few AI systems dominating across industries through M&A or partnership integrations.
Microsoft’s formidable position
Companies like Microsoft, which already have extensive access to data through products like Microsoft 365, will likely gain a considerable lead.
Microsoft’s competitive advantage comes from three key ingredients:
Microsoft 365 Apps: a large existing customer base (nearly 400 million Office 365 commercial users) to which Microsoft can sell Microsoft Copilot.
Unrivaled distribution channels that lower the barrier to adoption
Microsoft Graph: access to a vast corpus of organizational data, including emails, files, meetings, chats, and calendars.
This is the key differentiator. Prompts are processed through the ‘Microsoft Graph’ in a process they call ‘grounding’. This provides the prompt with business context so that users can get more relevant answers
This gives Microsoft a significant edge, as no other competitor possesses such comprehensive business context
Large Language Model (LLM): access to custom-made, proprietary OpenAI LLMs unavailable to competitors
Microsoft compared this to ‘driving the Ferrari’ vs ‘owning the Ferrari’
Combined with a 1-2 year head start (Microsoft first invested in OpenAI in 2019), Microsoft is a formidable contender in becoming the dominant enterprise AI system
It is is hard to see who has the capability to outmatch Microsoft in becoming the dominant enterprise AI system:
Salesforce is a likely contender as it owns the suite of front office applications (Sales, Service and Marketing Cloud), Slack (Teams competitor), Tableau (analytics), MuleSoft (data integration), and Salesforce Platform (app development)
Google has Google Workspace, but needs to do more work in integrating with other best of breed software so that it can create a cohesive enterprise AI system
SAP and Oracle both own both front office and back-office applications. However, they may be slow to innovate.
Conclusion
The future of enterprise software lies in AI-powered chat interfaces that can simplify user interaction, improve efficiency, and enable more powerful data-driven insights. As the industry moves towards a platform of platforms approach, competition will shift from UI to AI, and data access will become the key differentiator. The market will consolidate, with a few major AI systems emerging as dominant forces, shaping the way businesses interact with software and data.
Disclaimer: The content provided in this article is for informational purposes only and should not be considered as financial advice. The author is not a financial advisor, and this article does not constitute a recommendation to buy or sell any securities or make any investment decisions. Readers are encouraged to conduct their own research and consult with a qualified financial professional before making any financial decisions.