1995 and 2025: What is Different for Software Startups?

So some of us were around in 1995 to 2000 and working with startups, writing business plans and so forth. Compared to 2025, the software startup process looks really different: faster, cheaper, but maybe also with different metrics and processes for validating whether a market for the proposed product exists. 

And that matters, since some research suggests 35 percent of startups fail because customers did not actually want the product or solution. That might sound crazy, but it happens. I’ve lived it. 

 

source: Parallel 

So market validation is the process of testing whether real people in your target audience are willing to engage with or pay for your product idea, before you build it. These days, standard methods include data from landing pages, interviews, prototypes or paid pilots. If nobody signs up or pays, you’ve learned something valuable without wasting months of work.

Proof of concept metrics are different now, for example. 

Proof of Concept	1995	2025
What is "Proof"?	The team and the idea: A founding team with a strong pedigree and a convincing narrative about a revolutionary product.	Product-market fit: Concrete evidence that customers are using, loving, and paying for the solution without excessive sales or marketing effort.
Early Traction	Non-monetary milestones: Building the alpha/beta, acquiring a large number of free users, or securing a strategic partnership.	Monetary and retention milestones: $X in monthly recurring revenue, low logo/revenue churn rate, high user engagement (Daily Active Users), and successful paid pilots with defined use-cases.
The Ask	"We need money to build the product and launch the marketing campaign."	"We have proven we can acquire and retain customers efficiently. We need money to scale the proven go-to-market engine."
Defensibility	Proprietary tech: often a new database, networking protocol, or a patent-pending system.	Data and network effects: proprietary data sets, AI models, high switching costs, or platform-driven network effects (where the product gets better as more people use it).

Basically, market validation or proof of concept is the process of turning assumptions into evidence. 

Validation was more time-consuming, expensive and required larger teams back in 1995. 

Back then, during the dot-com bubble, money was not the problem, as odd as that seemed to me at the time. 

Back then, building software required $5 million and 20 people. Remember, this was before cloud computing and Amazon Web Services. We had to buy all our own “compute” platform before we could start. 

Faster, leaner compute also means founders can demonstrate traction much earlier than was possible in 1995. Prototypes can be created and launched much faster. 

Feature	1995	2025
Primary Method	Business plan and total addressable market: Focus on a compelling vision, "build it and they will come," and a massive Total Addressable Market (TAM) estimate.	Lean startup and minimum viable product: Focus on iterative, real-world customer experiments.
Cost of Validation	High: Required significant capital for dedicated servers, software licenses, in-house development, and formal market research.	Low (Near-Zero): Can use no-code/low-code tools (Webflow, Figma), cloud services (AWS/Azure/GCP), and affordable digital advertising.
Proof of Interest	Anecdotal: Focus groups, customer surveys, or Letters of Intent from large enterprises.	Behavioral and quantified: landing page conversion rate (waitlist sign-ups), pre-orders/paid pilots, and discovery interviews with 10–12 Ideal Customer Profile prospects.
Technology Focus	The product: The sheer ability to build complex, custom software was a barrier to entry and a selling point.	The value prop and distribution: The focus is on solving a specific, painful problem and proving a repeatable go-to-market (GTM) strategy.

Then it took 18 months to 24 months before founders had anything to show investors or customers.

Therefore founders needed a compelling enough story to raise millions based on a business plan alone. More to the point, they needed a compelling PowerPoint presentation. 

So everyone spent lots of time creating revenue projections (always up and to the right). And lots of us created some version of “Huge market of X size and we will get one percent.” 

Management team credentials were important, as everyone knew the hockey stick projections were illusory.

So successful founders were the ones who could raise money, based on past experience in related markets or simply because they had been successful before. It remains unclear if the best ideas won out. It is more correct to say the best-funded ideas often won. 

In 2025, with access to AWS, Google Cloud and Azure, everything is faster and cheaper. 

No-code tools and AI-assisted software development reduce validation time because creators can ship initial products fast and start learning from users in months, not quarters. 

“Traction” was important in 1995 as well, but that was largely “eyeballs” rather than revenue, in the consumer products space. Hence the focus on the audience: monthly active users, for example. 

Metric	1995	2025
Revenue Focus	Gross revenue/gross merchandise value: Total sales volume was the key (e-commerce transactions).	Recurring revenue: monthly/annual recurring revenue (MRR/ARR). High-quality, predictable revenue streams are prioritized.
Growth Metrics	Eyeball/User Count: Total users, website traffic, or simple month-over-month growth rate.	Net revenue retention (NRR): Measures revenue growth from existing customers (expansion minus churn). A >100% NRR is highly sought after.
Customer Value	N/A (or simple margin): Little focus on long-term value, as the business model was often ad- or transaction-based.	Unit economics: The relationship between Customer Lifetime Value (CLTV) and Customer Acquisition Cost (CAC). VCs typically want a CLTV:CAC ratio of 3:1 or better.
Financial Health	Cash on Hand: Focus on how long the runway lasted until the next raise (often called "burn rate").	Burn multiple: A measure of how much cash a company burns to generate $1 of new ARR. VCs look for low multiples (often <2x for B2B/SaaS) to prove efficient growth.

We might debate the relative importance of the various inputs or metrics. Founding teams and experience still seem to matter. But it might also arguably be the case that all the new tools could allow some startups to gain traction even in the absence of “founding team” experience. 

If a founder or founding team can go from concept to production-grade system with real paying customers in days to weeks to months, everything downstream can change as well. 

Success metrics could change from the number of validated users to actual revenue already being earned. 

When people can tell an AI engine what it is they want, and the AI can build the software, even fairly complex solutions, technical expertise in software development, access to capital or high-level domain experience might be less critical. 

Deep understanding of a particular process, coupled with sophisticated AI software development tools accessible using a natural language query process, might be enough, in many cases, to enable startups founded by people without all the traditional screening advantages venture capitalists look for.

If the Internet Collapsed "Distance," AI Collapses Time

If the core function of the internet is connectivity and the core value is the collapse of distance, then the core function of artificial intelligence is cognition and the core value is the collapse of time. 

If the internet makes physical location less important, AI makes complexity less important, reducing the time to derive insights. 

But both the internet and AI are going to disintermediate value chains, removing distribution functions and providers. 

Dimension	Internet	AI
Core Function	Connectivity: linking people, machines, data, and services across networks.	Cognition: performing tasks that require perception, reasoning, analysis, prediction, or decision support.
Primary Value Created	Eliminating distance: collapsing geography; enabling instantaneous communication and access.	Saving time: collapsing effort; automating, accelerating, or augmenting cognitive tasks.
Economic Logic	Reduces transaction and coordination costs associated with physical separation.	Reduces cognitive labor costs and enhances productivity by automating thinking tasks.
Primary Constraint	Bandwidth, latency, physical infrastructure (fiber, spectrum).	Quality of data, model capability, alignment with goals, compute.
Main Units of Scarcity	Transport capacity (Mbps/Gbps), access points (ports, routers), spectrum.	Compute, data quality, reasoning ability, task generalization.
User Experience Shift	From location-dependent to location-independent access.	From manual decision-making to automated or assisted decision-making.
Industrial Impact	Creates global digital markets; enables remote work, cloud services, platform economies.	Automates white-collar workflows; reshapes knowledge industries; introduces agentic systems.
Business Models	Subscription access, metered usage, advertising, platform marketplaces.	API usage, per-inference billing, embedded intelligence in existing software, agentic task fees.
Strategic Advantage	Owning the pipes, connectivity footprint, spectrum, and interconnection points.	Owning the models, data, workflows, and user attention for cognitive automation.
Regulatory Focus	Universal access, net neutrality, infrastructure competition.	Bias, transparency, safety, copyright, workforce displacement.
Transformation Pattern	Disintermediation of distance-dependent middlemen (retail → e-commerce; media → streaming).	Disintermediation of cognition-dependent middlemen (analysts, coordinators, support roles via agents).

So one way of understanding AI is to view it as a new form of infrastructure, as is the internet, as was electricity or railroads. In that view, potential over-investment happens because the new infrastructure has to be created, and not because of a mania or bubble over asset values that are illusory. 

That might temper some of the concern over AI asset valuations or investment magnitude, which can appear excessive in the near term, and might well be, in some instances. Such early over-investment tends to happen when a new general-purpose technology emerges, and especially when that GPT involves infrastructure.  

Historically, transformative infrastructure projects such as railroads experienced periods of perceived "over-investment," where excess capacity was common before widespread economic and societal adoption caught up. 

The U.S. railroad boom of the late 19th century and the electricity grid’s rollout involved capital surges, initial overbuilding, and even bankruptcies. However, over time, these investments generated foundational benefits, enabling entirely new industries and reshaping nations.

So although the superficial similarity between an irrational asset bubble and an infrastructure boom can exist, they are quite different. 

While a financial bubble features a disconnect between investment and credible returns, general-purpose infrastructure has long-term value, even if some amount of capital is misallocated. 

But that’s the issue right now: some see the infrastructure for a general-purpose technology being built; others see mostly speculation. It can be hard to tell the difference in the early going. 

Criterion	Productive Infrastructure	Speculative Excess
Cost-Benefit Analysis	Thorough, data-driven	Minimal or absent
Multiplier Effect	High, measurable output/wages	Weak, limited economic return
Demand Alignment	Supported by real user/market needs	Based on future hype, not evidence
Systemic Productivity	Positive spillovers	Neutral or negative impact
Asset Price Relationship	Aligned with long-term value	Driven by short-term speculation
Evaluation Rigor	Institutional, non-partisan reviews	Ad hoc, driven by momentum