Software Can't Keep Eating the World 🌎
For years, we've heard the rallying cry that "software is eating the world." The idea that every industry and sector will be disrupted and reshaped by software has become conventional wisdom in Silicon Valley. And the numbers seem to back it up - software and cloud services have appeared insatiable, devouring larger and larger portions of corporate spending. However, my take is that the idea isn’t as solid as it seems due to how much easier it is to build and replicate software. As software becomes easier to create, the age-old ‘build vs buy’ debate gains newfound relevance. Companies will retract from procuring external software that they could just build themselves. What it means for startups is that they may have diminishing moats.
Collapsing cost of software development
The very forces that have enabled software's takeover contains the seeds of an inevitable slowdown. Those forces? The relentlessly declining costs of computing power, cloud infrastructure, and software development itself.
The rise of abstraction layers and the ability to borrow infrastructure has shifted the competitive dynamics and lowered barriers to entry across many industries. Let's look at Banking-as-a-Service (BaaS) as an example. Fintechs and non-financial companies can now easily integrate banking services like checking accounts, loans, payments etc. by plugging into the APIs and regulatory pipes provided by firms like Synapse, Unit, Stripe, etc. This abstracts away the massive operational overhead of dealing with complex financial regulations and infrastructure. With BaaS, fintechs can then focus on creating innovative front-end products and experiences. Which is another way of saying that they are making large investments into sales/marketing (i.e high CAC).
Moore's Law has turned computing into a deflationary resource where mind-bending performance becomes exponentially cheaper every couple years. The cloud has allowed provisioning of compute, storage, and services on-demand with no upfront costs or overhead. And software itself has progressively lower marginal costs as open source and low-code/no-code tools proliferate. As these trends intensify, the economic incentives start to shift. Need a new enterprise app? Why pay SaaS fees indefinitely when you could develop a custom solution cheaply using serverless or reusable AI-assisted code components? Want to test an MVP? Just fire up cloud instance for a few bucks rather than signing an annual SaaS contract.
Long-term SaaS growth will continue slow down
Gartner forecasts that global enterprise software spending will reach over $1 trillion in 2024. SaaS spending has grown over 30% annually for the past 5 years, vastly outpacing overall GDP growth (~3%).
The steady decline in long-term SaaS growth rates for public companies, as depicted in the graph, provides tangible evidence that the astronomical rise of software and cloud services may be hitting a plateau. As the absolute scale of the industry increases, maintaining the year-over-year growth rates north of 30-40% is becoming unsustainable, even with the periodic cyclical upswings. SaaS growth cannot indefinitely outpace global GDP.
Accelerating Time to Market w/ AI
As the industry expands, achieving growth becomes increasingly arduous. Competition has intensified as companies vie for new bookings, exerting pressure on pricing structures and renewal rates. Moreover, the impact of AI on software will only exacerbate the challenge of sustaining growth. Democratization may be an overused term, but advances like low-code/no-code tools, open source software and AI assistants have unquestionably made it easier for less technical entrepreneurs, businesses and creators to build software-powered products and services. With AI assistance, spinning up a website, app or automating workflows becomes vastly more accessible to non-developers/non-technicals.
Applying new tech to old industries
So what’s next? If you ask me, I think there’s still work to be done in some of our oldest industries like Manufacturing, Industrials, Semiconductors, Hardware, Legal, Construction, Recruiting etc. Despite the influence of software in modern society, its proliferation across these older industries remains surprisingly bare bones. These sectors, collectively constitute a substantial portion of our global GDP. The question that’s been on my mind is “why haven’t these industries seen the light?” It’s like the cool kids (software engineers) are partying in one room with their MP3s, and the hardware folks are stuck in the basement still playing with their cassette tapes.
Software has definitely flirted with these domains, however the previous attempts often feel like they trying to force a square peg into a round hole — we’ve essentially misaligned software with the demands of these industries. Perhaps it’s time to take a cue from the strategic playbooks of companies like SpaceX, Tesla, Anduril, and Hadrian, where the integration of software into hardware is seamless and symbiotic. Rather than treating them as distinct entities, we should harmonize the two, marrying the world of atoms and the world of bits.
Here are are some of my favorite companies building at the intersection of software x hardware.
It’s time for Silicon Valley to rekindle its relationship with hard tech, welcoming it with open arms. Like greeting an old friend. HardTech isn’t just a nod to the past, it’s an opportunity to unlock untapped potential and address real-world challenges with tangible solutions.