Why AI is Attracting VC Backing for Software Development

If scale is our measure, we must concede that it takes money to make money. The multiplier effect yields greater rewards when you begin with a substantial base of wealth. That’s why we’ve seen scads of moolah pursue the evasive holy grail of generative AI, which promises prodigious, if not unprecedented, returns on investment.

So far, though, the commercial promise of generative AI remains largely unrealized. Alas, it might remain that way, at least in relation to exuberant market expectations.

Don’t get me wrong. I’m not saying that AI is without its uses. In fact, it’s conceivable that AI will eventually become practically ubiquitous, embedded in nearly everything we see and touch. At that point, presuming it happens, AI’s pervasive familiarity will gradually strip away its exoticism and novelty. AI will become something we take for granted, a banal presence, like the computer or the smartphone today, though obviously not as tangible in the physical sense as either of those technological mainstays.

That brings us to another salient point. Essentially, AI will always remain an enabler, a facilitator, a mechanism applied for the purposes of achieving a desired outcome. Companies and consumers will not adopt AI because it’s AI — which, shorn of context, does nothing — but because AI will assist in completing tasks and achieving outcomes more expeditiously and efficiently than conventional means.

Long ago, pragmatism taught us that a thing is what it does. All these decades later, we still struggle to internalize and apply that lesson. Now is probably a good time for us to recall the abiding value of pragmatic functionality.

AI will only attain commercial value through its ability to provide tangible outcomes. Nobody pays for AI as a thing in itself, as a fascinating novelty or an innovative curio, but they might pay for AI if it helps them dispense with daily drudgery, thus saving time; or if it expedites processes and workflows more efficiency and productively than teams of humans are capable of doing. The intricacies of technical mechanisms might enamor geeks, but arcane charms don’t pay the bills. Results and business outcomes are the foundation on which tech mansions are built.

Software Development: The AI Headpin?

In an article yesterday, Reuters noted that “two years after the launch of ChatGPT, return on investment in generative AI has been elusive, but one area stands out: software development.”

Why is that? Why should AI deliver ROI in software development, but not in some of the more grandiose and whimsical realms — life coaching, for example — that venture capitalists touted breathlessly not that long ago?

There’s more than one reason, of course, but we can state conclusively that the perceived and real value of an AI life coach is nebulous, whereas the value of AI-enabled “vibe coding” and software development is increasingly tangible. One can apprehend that value in dollars and cents, manifested on the bottom line. Such uses have undeniable monetary value, which means they are amenable to the mathematics of value propositions and return-on-investment calculations. To put it more bluntly, these applications speak the numeric language of CFOs.

With that in mind, here’s a helpful excerpt from the Reuters article:

So-called code generation or “code-gen” startups are commanding sky-high valuations as corporate boardrooms look to use AI to aid, and sometimes to replace, expensive human software engineers.
Cursor, a code generation startup based in San Francisco that can suggest and complete lines of code and write whole sections of code autonomously, raised $900 million at a $10 billion valuation in May from a who’s who list of tech investors, including Thrive Capital, Andreessen Horowitz and Accel.
Windsurf, a Mountain View-based startup behind the popular AI coding tool Codeium, attracted the attention of ChatGPT maker OpenAI, which is now in talks to acquire the company for $3 billion, sources familiar with the matter told Reuters.
Its tool is known for translating plain English commands into code, sometimes called “vibe coding,” which allows people with no knowledge of computer languages to write software. OpenAI and Windsurf declined to comment on the acquisition.

Feeling the Vibes

In the above excerpt, Reuters takes a cursory swipe at defining vibe coding, but we should define it more substantively if we are to grasp the complementary value that AI adds to software development.

As a prologue, I apologize to those of you already familiar with the nuances of vibe coding. Some of you doubtless are more conversant with the term than I, but I would rather not assume that every reader has added this relatively shiny new term to his or her lexicon.

Vibe coding denotes a programming approach in which developers rely on intuition, feel, and informal understanding rather than on strictly prescriptive methodologies and detailed planning. Notable attributes of vibe coding include writing code on the based on what feels right; iterative discovery of what works and what doesn’t; an emphasis on momentum and creative flow; a minimal formal structure (within reason); and a focus on pattern recognition rather than on an observance of hard-and-fast rules. The of it as software development as artistic expression. Perhaps Salvador Dali would approve.

Let’s circumvent the debate on whether this approach is good or bad, right or wrong. People will disagree on so-called vibe coding, as on sundry other topics. For the record, I will acknowledge that many cognoscenti suggest that the new approach can be effective for seasoned developers working on creative projects and prototypes. Conversely, the general view is that vibe coding is less productive in scenarios that involve large teams, mission-critical systems, or projects requiring extensive collaboration and maintenance.

I should also say that the case for and against vibe coding is still being debated before judge and jury, and a definitive verdict has yet to be delivered.

As for AI’s contribution, the supposition is that AI tools might significantly enhance the capabilities and strengths of vibe coding while mitigating perceived weaknesses. How? The thinking is that AI will infuse vibe coding with real-time idea translation (turning fuzzy concepts into functional code); rapid prototyping; pattern completion; experimental assistance; context-aware suggestions, and a lot of under-the-covers minutiae and drudgery that is of comparatively lesser value in the overall context of the project.

It’s early days for AI and vibe coding, so the usual caveats apply. Mileage may vary, and so on and so forth. The theory, however, is that AI amplifies, rather than inspires and constructs, the purported creativity of vibe coding.

VCs Play the Long Game

Venture capitalists are buying into the narrative. Many VCs, including those in offices at some of the shiniest shingles on Sand Hill Road, believe AI and vibe coding will make more money than the combination of chocolate and peanut butter. In the Reuters article, two accomplished VCs — Martin Casado of Andreessen Horowitz and Scott Raney of Redpoint Ventures — are quoted. (Quick personal aside: In my prior professional incarnations, I’ve benefited from informative and stimulating conversations with both men.)

VCs point to a vast addressable market, stressing that software development encompasses hundreds of billions of dollars in spending annually. The belief is that any tool that can significantly accelerate development cycles will find an eager paying audience. Further, VCs believe that developer productivity can be massively boosted and that software development itself can be democratized; that is, opened to a broader base of users. For enterprises and other organizations, especially those employing large numbers of conventional developers, the cost savings from AI-infused vibe coding could be considerable.

In life, as in business, I have learned that what remains unsaid often is at least as important as what is expressly said. For the most part, people refrain from saying the quiet part out loud, but they’re thinking it all the same. You know what they’re thinking, but not necessarily saying, when it comes to AI-assisted vibe coding, don’t you?

What these in-the-know people are thinking is that AI-goosed coding will potentially displace software-development jobs by the boatload, but they suppress the urge to express those thoughts verbally. That’s partly because employers, the customers of the AI purveyors, don’t want employees to know that their end is nigh, and partly because software developers will initially be using AI to augment their capabilities and expedite human tasks. Among those who want to keep their jobs, few wish to train and enable their successors.

AI coding startups need developers to adopt their tools to validate effectiveness and spark early adoption. While augmentation is part of the value proposition, and the prevalent narrative theme in the market today, a larger revenue lift will derive from persuading CFOs and other C-level executives that AI tools will enable leaner, less costly operations. Before AI purveyors reach that stage, however, they’ll promote their tools on the politically acceptable grounds of productivity augmentation, through which developers achieve greater efficiencies and faster code delivery.

What this scenario suggests is that early adopters of AI tools, the software developers, are helping to refine tools that could eventually put them, or their subordinates, out of work. Many developers today likely have sensed that such a dynamic is in play, but some have decided that they’re better off playing along and trying to adapt professionally to the changing circumstances. Eventually, the thinking goes, this stuff will be everywhere, and the entire job market will be reconfigured. Why hang on to the past when the past is about to be relegated to history texts and sealed behind a glass case in a museum?

Meanwhile, VCs are playing the long game, foreseeing a timeline in which AI customers advance from augmenting the productivity and efficiency of their developers before figuring out how they can go even faster while smaller development teams.

No Guarantee, Only Probabilities

That’s not to say that the plan will come together perfectly in a confluence of exquisite timing and deft execution. You need expertise, skill, resources, timing, and luck — not always in equal complements or in that order — to hit the jackpot in technology, and many contingencies can derail the money express. Still, I think the big-picture intent is clear.

Among the risks, there’s always the threat that somebody with more resources, pockets stuffed with more money and also in possession of the requisite infrastructure, will leave you in the dust. As the Reuters piece says:

But because most are built on AI foundation models developed elsewhere, such as OpenAI, Anthropic, or DeepSeek, their costs per query are also growing, and none are yet profitable.
They’re also at risk of being disrupted by Google, Microsoft and OpenAI, which all announced new code-gen products in May, and Anthropic is also working on one as well, two sources familiar with the matter told Reuters.
The rapid growth of these startups is coming despite competing on big tech's home turf. Microsoft’s GitHub Copilot, launched in 2021 and considered code-gen’s dominant player, grew to over $500 million in revenue last year, according to a source familiar with the matter.

In addition to the risk that industry giants could swamp the market with their own AI-powered code-generation tools, VC-funded startups will have to compete for survival against a growing number of peers. To survive an inevitable shakeout, startups are blazing one of two general paths: a blitzkrieg land grab, taking as much early market share as possible, or targeted and defensible differentiation that might result either in standalone success or, more likely, exit through a roll-up acquisition by a larger player.

Startups are already accruing significant revenue. To wit (again from Reuters):

Cursor, with just 60 employees, went from zero to $100 million in recurring revenue by January 2025, less than two years since its launch. Windsurf, founded in 2021, launched its code generation product in November 2024 and is already bringing in $50 million in annualized revenue, according to a source familiar with the company.
But both startups operate with negative gross margins, meaning they spend more than they make, according to four investor sources familiar with their operations.
“The prices people are paying for coding assistants are going to get more expensive,” Quinn Slack, CEO at coding startup Sourcegraph, told Reuters.
To make the higher cost an easier pill to swallow for customers, Sourcegraph is now offering a drop-down menu to let users choose which models they want to work with, from open source models such as DeepSeek to the most advanced reasoning models from Anthropic and OpenAI so they can opt for cheaper models for basic questions.
Both Cursor and Windsurf are led by recent MIT graduates in their twenties, and exemplify the gold rush era of the AI startup scene. “I haven’t seen people working this hard since the first Internet boom,” said Martin Casado, a general partner at Andreessen Horowitz, an investor in Anysphere, the company behind Cursor.

To guard against irrelevance that might result from usurpation by AI and cloud giants — the former represented by OpenAI and Anthropic, the latter by AWS, Microsoft, and Google — some startups are building their own models. That’s costly, but it could pay dividends if they do it right and achieve the sort of differentiation that is appreciated by paying customers. (The word differentiation is thrown around indiscriminately, like a beachball at a well-lubricated pool party. There’s purely technical differentiation, which often matters only to its engineering progenitors, and there’s true market differentiation, which wins the affections and wallets of patrons. We refer to the latter variety here.)

In the Reuters piece, Scott Raney makes that very point:

“In many cases, it's less about who's got the best technology – it’s about who is going to make the best use of that technology, and who's going to be able to sell their products better than others,” said Scott Raney, managing director at Redpoint Ventures, whose firm invested in Sourcegraph and Poolside, a software development startup that’s building its own AI foundation model.

Is AI-infused vibe coding already displacing developer jobs? Maybe.

A commentary at Marketplace, citing a Bloomberg report, notes that Microsoft last month laid off about three percent of its workforce, totaling about 6,000 employees. A large portion of those culled employees were reportedly software engineers.

In reducing its software-engineering headcount, Microsoft might first have eaten its own AI dogfood. During the same month, the company announced a suite of AI products that included what it described as an “asynchronous coding agent” that is allegedly capable of fixing bugs, adding features, and performing other programming tasks.

We don’t know how many software jobs will be displaced by AI. For one thing, nobody is keeping a systematic count, and that’s partly because employers and AI companies aren’t reporting such data. You can understand their reticence.

From the piece at Marketplace:

“What happens to coding and software engineering is a bellwether,” said Molly Kinder, a fellow at the Brookings Institution.
She said software engineering is the first occupation to use this technology en masse. And if AI is pretty good at writing a college history paper or making an image of the pope in a funny coat, it’s really good at programming.
. . ..
Kinder from Brookings said that whole idea of software engineering as the stable career of the present and future is so last decade.
“I recently looked back at the 2015 Obama White House AI and automation report. One of their recommendations is, everyone should learn coding,” she said. “And you know, you cringe reading it now.”

We move fast through time, and the scenery and circumstances are subject to occasionally wrenching change. The present is just a snapshot, and even the present has many dimensions, depending on one’s vantage point.

For now, however, I’d say circumstantial evidence suggests that the hypothesis around AI as an enabler of vibe coding is being corroborated at web giants and other large software developers. AI obviously isn’t about to be wielded to eliminate every developer job, and it’s possible that market and labor adaptation will generate a commensurate number of compensatory jobs.

Nonetheless, corporate application of AI does appear to be displacing software jobs, especially entry-level or junior positions. The architects are safe and secure, it seems, because somebody has to give purpose to the automated drudgery.

Generally, I’m coming to the view that AI will do more than productivity augmentation — what a term, by the way — as it gains traction.

At one time, as alluded to earlier, guidance counselors, politicians, and pundits advised young people to code. I wouldn’t give that advice now. Instead, I think the key to self-preservation in the job market involves the development and utilization of what we call soft skills, which the Oxford English Dictionary helpfully defines as “personal attributes that enable someone to interact effectively and harmoniously with other people.”

AI bots can’t do those sorts of jobs, and I suspect they won’t be able to do them in your lifetime. AI bots can’t get on a plane, attend trade shows and conferences, visit and meaningfully interact with a valued client, or demonstrate empathy and concern in person or on a video call. Nor can they react in real time to subtle verbal or non-verbal cues, which are essential capabilities and attributes in any negotiation or meaningful business discussion. Besides, people are naturally averse to negotiating with a bot.

If you’re doing work in isolation from other people, and your job involves repetitive processes and regimented workflow, you might be in the AI danger zone. A focus on the interpersonal, on serving as bridge or conduit between other key stakeholders, is the path I prescribe for those who wish to maintain economic relevance and career longevity.