Salesforce's earnings last week were poorly received by the market.
The CRM software giant issued its worst quarterly revenue growth forecast ever, and its shares fell 18% after the numbers were released in after-hours trading on May 30. This represents a massive $48.5 billion loss from its market cap.
The same thing happened with human resources software group Workday last week, whose shares fell 20% after it cut its outlook for next year.
What is going on here?
Think back to August 2011, when Marc Andreessen, arguably the most influential venture capital investor of the Silicon Valley era, published an essay in the Wall Street Journal titled, “Why Software is Ruling the World.”
Andreessen argued that software companies are disrupting and transforming traditional industries.
“I expect that over the next decade many more industries will be disrupted by software, and in most cases that disruption will be driven by the world's leading Silicon Valley startups,” he wrote.
Eat or be eaten
The essay highlighted how the rise of software and internet companies like Google and Amazon is upending established businesses across many sectors, and Andreessen predicted that this trend will accelerate, with software continuing to “eat the world” by disrupting and reshaping industries not previously affected by the technology.
More than a decade later, Andreessen's prediction has proven to be remarkably accurate. Many of the world's largest and most valuable companies today, such as Microsoft, Amazon, and Google, are primarily software-driven businesses that have crushed traditional businesses in media, entertainment, publishing, recruitment, advertising, and retail.
New Menu
The Q1 2024 earnings season has provided strong evidence that a new era of gourmet food is upon us: AI is devouring software, and more specifically, application software.
Some definitions: Application software performs a specific business function, such as design, video editing, or tracking customer interactions, as in the case of Salesforce.
This is distinct from infrastructure and enterprise software, which provides core functions for a company's day-to-day operations (such as management reporting, supporting production operations, and back-office functions).
And, an important note: some application providers that are early adopters of AI built into the core of their software, as Microsoft did with Copilot, will survive and thrive.
There are many possible explanations for this indigestion. But first, kudos to Peter Singlehurst, head of the private companies team at Baillie Gifford, who last year nailed it in an article titled “The venture-backed, low-capitalization software business is dead.” May he rest in peace..
Reasons for feeling down
Here are some of the reasons I can think of for this reversal of fortune.
- Application software providers are seeing spending decline due to a slowdown in the U.S. and global economies caused by high interest rates.
- Their clients are shifting spending away from application software and towards AI.
- These companies are restructuring to counter disruption from new “AI-at-core” competitors and are facing rising costs.
- We also see limited and slow adoption of unique “AI as an add-on” applications.
- Application software companies have long released new versions of their software that add significantly to their costs while providing little benefit to their clients.
- Investors are weary of stock-based compensation plans that give all the profits back to management while shareholders bear the risks, something I wrote about previously in a column called “Who's Paying Who?”
Regarding the last point, The Wall Street Journal Salesforce pays the movie actor Matthew McConaughey to be a “creative advisor and TV publicist,” and the less-than-generous description of the relationship suggests his unspoken role was to make CEO Marc Benioff look good at parties.
Change the code flash
I want to focus on points 4 and 5 above. The key to these points is that too many application software companies treat AI as an “add-on” that they tack on to their core code, which is standard sequential code. They do this because they are afraid to change the core architecture that their software is based on to an AI-based core.
I think this timidity is likely to lead to confusion. “Sequential legacy code” cannot learn about the specific enterprises that are using it to tailor or customize the software, but “AI at the Core” applications can and will. In the future, such software will be able to rewrite the code and re-adapt its offerings as the underlying enterprise's circumstances change.
History may be repeating itself (or at least rhyming) here, as a young Marc Andreessen re-emerged in the 1990s to develop the Netscape browser (remember those?), which helped move the world from writing code against standard Windows APIs to internet-based APIs.
This was a move from a more sequential programming paradigm to a more asynchronous, event-driven approach. Without worrying about the technical details, this move is very similar to how we might move from traditional sequential code to AI-based code generation.
Just as internet-based APIs abstract away low-level details and provide a high-level interface, AI-based code generation aims to abstract away the complexities of writing code by leveraging machine learning models to generate code based on natural language input or other high-level specifications.
Internet-based APIs often involve asynchronous communication and event-driven programming, where code responds to external events or data. Similarly, AI-based code generation can enable a more reactive, event-driven approach, where code generates or self-adapts in response to changing requirements or inputs.
Specialists and generalists
This could have major implications for traditional application software providers.
Here's why.
To remain competitive, they would have to completely redesign their applications, but doing so would likely destroy a key barrier to entry: the stickiness of a user base that is often loyal to an application because they know exactly how to use it.
Changing the way software works seems like it would likely result in a hostile reaction from old users. Remember when Coca-Cola launched their new flavor, New Coke? The dislike was immense, and New Coke quickly became Ex-Coke.
Luckily, professional technology investors know this and are watching to see what happens.
None of the technology investment trusts listed on the London Stock Exchange have significant exposure to application software companies. If you have time, take a closer look at recent factsheets on Allianz Technology Trust, Polar Capital Technology Trust, Scottish Mortgage and Manchester & London. None of them have application software stocks in their top 10 (except Microsoft).
However, you should do your due diligence on generalist portfolio managers: you may find that many of these managers who don’t have the time or skills to get deep into the field hold key positions in these companies.
For anyone with a large chunk of their client's money invested in Salesforce, last week's stock price crash served as a painful lesson.
Secret Tech Investor is a seasoned professional who has been managing technology assets for over 20 years.
