AI is eating software (and creating a parallel labour force)
How this new dynamic is changing the venture capital industry and startups
In our latest conversation for Giant Ideas, we asked Babylon founder Ali Parsa, who recently launched an AI health assistant, his predictions for the future of AI. He responded: “In the Middle East, we say, ‘If you want God to laugh, you give her your two-year plan!’” Indeed, it is clear that AI is moving extremely quickly, so we humbly proffer any predictions, but one thing that is fast becoming clear that AI is not just a technology platform shift but a paradigmatic shift in how human labour relates to our economy.
The history of software
For the last twenty years, software has often appeared as magic: making the impossible possible. The first era of Internet innovation - the so-called “dot com” boom - was mostly about bringing services online (Amazon, Pets, Webvan), as well as building web portals (Yahoo, AOL, & Google). The more recent wave, from 2007 to 2020, saw many entrepreneurs create an online overlay on our offline reality, disrupting legacy industries like transport (Uber), payments (PayPal/Stripe), hospitality (Airbnb) and dating (Tinder). They also built infrastructure and tools, as Snowflake and Datadog did, for legacy businesses to ride the digital wave. This wave of innovation also led to some of the most outstanding investment returns ever. Now AI is eating software, eroding the fundamental economic value and creating a new era that looks ripe to disrupt inflationary industries, such as healthcare, education, and construction.
Software is no longer enough to create value
The past decade saw an explosion of SaaS companies, estimated at close to 35,000 in 2024. In the future, products such as these will still be built, but they will increasingly be launched by bootstrapped founders building lifestyle businesses. It is difficult to see many of these companies delivering venture-type returns because the barriers to entry have been lowered massively. Until recently, software engineers were most startups' main cost (and value). Fast-forward to today, and tools such as Microsoft's GitHub Copilot, powered by generative AI models, can now write significant portions of code.
What was once a deep moat around general software companies—the specialised expertise and investment required to build —is no longer. Therefore, the ability to capture value, market share, and charge premiums will diminish over time. Mature software companies like Google, Facebook, and Salesforce are still growing rapidly, benefitting noticeably from their data, computing, and distribution advantages, but up-and-coming startups must be creative in their approaches to create and capture new value.
AI as a parallel labour force
AI is now creating a parallel labour force - altering the economics of labour by reducing the cost of delivery. In time, AI should be a deflationary force as it reduces the cost of inputs, leading to lower prices. The price of a television set in the United States fell by 98% between 1997 and 2022, and computer and software items today are 74% cheaper than 25 years ago, according to the U.S. Bureau of Labor Statistics. AI is now reducing the cost of technological applications. But AI's real promise is to slash the cost of labour-intensive services such as healthcare, engineering, or education. Whether it’s tutoring, maintenance, or nursing, AI is transforming our relationship with labour, just as the internet transformed our relationship with the offline world.
Our latest investments speak to this trend: from building an AI-native operating system for hospitals, to an AI copilot that enables more efficient advanced manufacturing, to building autonomous robotic construction solutions. These are all powerful examples of a parallel labour force in action and a new value proposition around augmenting or replacing the labour currently involved.
From enablement to full stack
Over time, we expect these startups to use AI to capture more of the value chain and automate more labour. More startups will vertically integrate to offer a complete service, thereby capturing more of the value chain. The emergence of this type of parallel labour force may change how startups create and capture value. This is best elucidated through examples.
For instance, consider a customer experience startup. In the software era, this startup might have offered software to a call centre company that provides customer service support to Fortune 500 corporations to optimise its call centre operations. Now, a startup is more likely to build a full-stack startup that offers customer service support directly to Fortune 500 companies, bypassing the call centre intermediary. Such a startup would essentially build an asset-light version of an AI-first “call centre” from the ground up, leveraging AI voice and text applications to offer a parallel labour force (customer service in this case) as a service.
As discussed in a previous piece, all these startups will need a unique data set and distribution advantage to compete. That’s the starter fuel you need for your moat for AI. They will also need access to vast computing power. These startups likely won’t be as capital-efficient as traditional software startups, although AI usage costs are falling quickly. They may have lower margins and need to create new playbooks for scaling. However, the potential market size for these startups will arguably be larger because they can pursue the entire value chain by delivering a parallel labour force while remaining, at their core, technology companies. And that’s exciting.
How Giant is expanding its investment aperture
Finally, software’s changing economics, diminishing defensibility, and limitations for meaningfully solving the climate crisis have led us to widen our aperture to more deep tech opportunities. Harking back to the genesis of venture capital and the “silicon” chip days, deep tech often requires a multidisciplinary team, hardware and software integration, and fundamental advances in physical and biological systems. There may be more technology risk, and the development cycle can be longer, but there is also a natural barrier to entry when major milestones are achieved. The upside potential can be significant - both for impact and returns. We’re actively looking at breakthrough technologies, and we are building businesses at the frontier of scientific discovery and engineering feats.
I’m excited and also uncertain about how the world will look a decade from now. At Giant, we closely monitor how AI is transforming company building, enterprise value creation, impact, and more. We see real potential for AI to transform the historically inflationary industries of healthcare and education, as well as to transform frontier science. We remain motivated to positively shape the future by building and backing the companies that matter and are excited about the current cohort of entrepreneurs using AI’s magic capability to build that future.
For more insights, check out our latest Giant Ideas episodes with Huffington Post & Thrive Global Founder Arianna Huffington, Bridgewater Associates Founder Ray Dalio, or Former Sony Entertainment CEO Michael Lynton.