For centuries, the great goal of science has been alchemy: turning lead into gold. This is actually possible in nuclear physics, and is called transmutation. If you bombard mercury with neutrons in a nuclear reactor or particle accelerator, you get… gold.
Today, thanks to AI, we may be witnessing the greatest transformation in history: software becomes labor. This is the new E=MC2. Capital buys coffee, engineers, and GPUs. The result is code that takes on the role of labor. This expands existing software markets and creates many new software markets where “per seat” pricing has not had much success.
Historically, much of the software has digitized offline forms of storage, stored them in databases, and taken advantage of the speed of digitized, networked media to provide an accessible, authorizable front end to end users who may not know SQL.
PeopleSoft, and later Workday, digitized the HR filing cabinet.
Zendesk has digitized the old-fashioned support “ticket.”
Quicken and Quickbooks have digitized old-fashioned ledgers.
Epic and Cerner have digitized health records.
Salesforce has digitized the Rolodex and the chalkboard pipeline.
Email has gone digital and is now mail!
Instead of Bill from HR and Sally from Accounting literally “pulling” files out of a filing cabinet, it's William and Sarah from IT making sure everyone in HR and Accounting is properly provisioned and has access to the systems — except the number of people in Accounting is the same as the number of people in HR.
But now, and this is a big change, the “users” of a digitized filing cabinet don't have to be human.
Let’s take a look at Zendesk’s pricing.
There is now a huge opportunity for all the “digital file cabinet” companies like Intuit, Workday, Zendesk, etc. to charge for “quasi-humans” who replace or supplement the humans who previously accessed and acted on records. This could necessitate a massive change to traditional “per seat” pricing models, as companies will need fewer seats (if any at all) as systems take actions independently or increase human productivity by 10x.
Workday can charge for a conversation with an HR rep. Intuit can charge for sending a collection letter for a past due account receivable. Epic can charge for a post-meeting check-in. Salesforce can charge for a sale.
To be clear, this does not mean the end of white-collar jobs. Rather, AI will likely create new “AI jobs” that would not be feasible with human costs or intermittent demand. There's a famous saying in economics: “The cure for high prices is higher prices.” That is, when the price of a good rises, more manufacturers decide to produce that good (or existing manufacturers produce more), so supply increases and the price falls. But for skilled labor, the wait is simply too long. If nurses' salaries rose five-fold tomorrow, it might not make any noticeable difference to tomorrow's supply (some nurses who left might even undo their retirements!), but it might result in more nurses entering nursing school and graduating in three years. Three years. What happens if there's a surge in nurses tomorrow?
For other professions, training is “on the job” and it is impossible to align training hours with the ups and downs of business. Demand for mortgage brokers soared in 2021 as mortgage rates fell, but it ended abruptly when rates rose. Delta Airlines needed perhaps tens of thousands of trained support personnel following the CrowdStrike disaster. The fifth fastest growing job in the US over the past 20 years is “compliance officer,” but even here training (not just wages!) creates shortages. AI always shows up for work, is quickly trainable, and is happy to fill the “market void” that occurs when demand for highly skilled jobs is intermittent or temporary. Why go to school or spend years learning something if your skills are only needed twice a year? AI doesn't have this problem.
From a venture capital perspective, almost every company has a compelling “why now” (if it's such a good idea, why hasn't it worked out before?), but I like to think that there have been three eras of cloud software.
In the original cloud era (~1999-2007), the market was large enough for pure software to largely replace on-premise software. This is when players like Salesforce, NetSuite, Veeva, Hubspot, etc. emerged. In the financial services-enabled cloud era (~2010-present), the market was large enough with built-in payments, lending, and other “fintech” add-ons. For example, the restaurant software market was small, but the restaurant “software + payments” market was large, hence Toast (founded in 2012). The HVAC contractor software market was small, but the “software + payments” market was large, hence ServiceTitan (founded in 2012), etc. In the “AI-enabled outcomes” cloud era (present – ), the software market was small because labor was the primary cost and software needs (digital filing cabinets) were relatively small or generic. Many professional software providers only provide Excel and Word, rather than specialized tools.
This is not to say that AI will only be successful in “brand new” areas. But as is often the case, in the battle between distribution and innovation, distribution is the default winner, and so are the incumbent software companies. Especially now, in 2024, AI is a top priority for almost every software company and end customer. In 2007, most CEOs thought the iPhone was stupid (no keyboard!) and Blackberry was better. In 1996, most retail CEOs thought the Internet was a toy and a passing fad and that people would never buy things from a web browser. So new companies could fill the void. In 2024, it’s almost impossible to find a CEO who thinks AI is a bad idea.
AI software companies have essentially three different origins and outcomes.
AI tools that run on top of existing software (e.g. automated meeting notes for Zoom meetings) AI tools that run on top of existing software and may replace that existing software (e.g. meeting notes for Zoom meetings…the company will build video conferencing and encourage you to get off Zoom) AI tools that transform the workforce — an entirely new category, something that hasn’t been impacted by software at all before (e.g. software running meetings for you)
Platform shifts have always enabled the first two (an internet version of X, a mobile version of X, a cloud version of X). But what's most exciting about the AI revolution is that the enterprise software market, which seems large with $300 billion in annual spending, is tiny compared to the trillions of dollars per year white-collar labor market. This is why many of the fastest-growing companies we see have been “known unknowns” – taking existing expensive services and offering low-cost products (created by AI) to the masses.
And we are at the beginning of software consuming and augmenting labor.
