Venture Bytes #117: ‘Vibe Coding’ is Silicon Valley’s Next Act

Vibe coding, a term coined by OpenAI’s cofounder Andrej Karpathy, which refers to using AI to generate code via natural language prompts, is emerging as the next frontier in software development. OpenAI CEO Sam Altman recently predicted that software engineering will be “very different by the end of 2025.” Not long after, Mark Zuckerberg hinted that AI is poised to take over tasks traditionally handled by midlevel engineers at Meta.

‍
Vibe coding is already here. AI now generates 41% of all code, with 256 billion lines written in 2024 alone, according to data from GitHub. That’s a staggering figure — and this is just the beginning. Also, 92% of US-based developers are already using AI coding tools both in and outside of work. With each iteration, AI is improving at an exponential rate, and developers are rapidly adapting.

It’s not just individual developers driving this shift; enterprises are embracing AI-powered coding at scale. A survey of founders from Y Combinator’s current batch revealed that 25% of them have AI writing 95% of their codebases, according to YC managing partner Jared Friedman. Moreover, more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers, according to CEO Sundar Pichai. This highlights how large organizations are also adopting vibe coding, amplifying its industry-wide influence.

‍
The implications are massive. Startups can build at unprecedented speeds, reducing the time from idea to MVP from months to weeks. Non-technical founders can ship software with minimal human coding, reshaping what it means to be a “technical” entrepreneur. On the flip side, as AI coding adoption grows, the sheer volume of software produced will skyrocket, raising questions about code quality, maintainability, and security risks.

‍
The rise of AI-driven coding won’t play out in a straight line. Instead, we expect a bifurcation in software development with developing simple apps becoming nearly effortless. Less complex applications will be built in a few prompts, eliminating the need for large engineering teams. This will likely lead to a surge in AI- generated micro-SaaS tools, automation scripts, and simple web apps—many of which can be “one-shotted” with minimal human input. However, complex applications will become harder to build. As AI takes over code generation, developers will increasingly lose deep understanding of their own codebases. Over-reliance on AI will lead to diminishing returns, especially for large-scale systems that require architecture, debugging, and long-term maintainability.

‍
Just as AWS abstracted away infrastructure headaches for startups, AI-powered coding tools could abstract away much of the complexity of development. This shift could create massive new markets in AI-first development environments. Accordingly, VCs and developers are more excited than ever about AI-powered coding. Over the past 12 months, startups in this space — including Bolt.new, Codeium, Anysphere, Lovable, and Magic—have raised hundreds of millions of dollars in funding.

Among the startups driving this revolution, Codeium emerges as a leader and a prime investment opportunity. Its AI-powered platform supports over 70 programming languages and integrates with more than 40 IDEs, offering unmatched versatility. In August 2024, Codeium secured $150 million in a Series C round at a $1.25 billion valuation, led by General Catalyst, Kleiner Perkins, and Greenoaks, and by February 2025, it reached $40 million in ARR, with talks for a new round at a $2.85 billion valuation. Codeium has onboarded over 1,000 businesses, including Anduril, Zillow, and Dell.

Anysphere also has quickly become a standout player in AI- powered coding. In just 12 months, Anysphere hit $100 million in ARR, with its valuation soaring from $400 million to $2.6 billion in five months. Anysphere has secured 30,000+ developers and enterprise customers like Shopify, OpenAI, and Instacart, making it one of the fastest-scaling AI coding startups.

California-based Replit is another fast-growing startup. Its cloud- based platform has fueled rapid growth, expanding its user base from 10 million to 22.5 million developers between December 2021 and April 2023. The company has strategic partnerships with Google Cloud and Anthropic, providing top-tier infrastructure and AI capabilities. Serving 22+ million developers, from solo programmers to enterprise teams, Replit is well-positioned to drive the future of scalable       d coding, making it compelling bet for investors.

AI Could Be Your Lifeline in the Next Crisis

‍

"In the critical minutes that separate life from death, Artificial Intelligence (AI) is emerging as one of the most powerful tools to augment emergency services. Historically, emergency services have been reactionary — responding once incidents are reported. AI, however, is flipping that paradigm. Using machine learning, computer vision, and predictive analytics, emergency services are now anticipating crises, triaging faster, and optimizing responses like never before.

‍
The consequences of inaction or late action are profound. The catastrophic fires that devastated Greater Los Angeles ravaged more than 40,000 acres, destroyed upwards of 12,300 structures, and compelled the evacuation of as many as 200,000 residents, according to the US Census Bureau. In response, over 9,000 firefighters were mobilized, yet by the time the flames were fin ly contained, the total damage had already eclipsed $150 billion.

This disaster, however, is not an anomaly; it is merely one example in a broader pattern of ongoing crises. The US continues to contend with an array of persistent emergencies, including mass shootings, car accidents, and heart attacks, among others. In 2023 alone, there were 620+ mass shootings—an average of nearly two per day, per Gun Violence Archive. Additionally, over 6 million passenger car accidents are reported annually, according to Darrigo and Diaz, while approximately 805,000 Americans suffer from heart attacks each year, as documented by the US Department of Health. Further, there is a dire lack of public health workforce with projections estimating a need for the US public health workforce to grow by 80%, according to de Beaumont Foundation, which advan  builds partnerships, and strengthens public health.

The US emergency infrastructure is buckling under massive demand. On New Year’s Day 2025, the nation experienced an unprecedented surge in 911 calls, reaching a peak of 690 emergencies per second at 1 AM ET, as noted by Michael Martin, cofounder of RapidSOS. The year 2024 marked a record high of 2 billion 911 calls, underscoring the mounting strain on the emergency response infrastructure. Despite advancements in technology, the underlying 911 system remains reliant on outdated structures that hinder efficient communication during crises.

‍
The ubiquity of connected devices presents an opportunity for emergency response. According to Pew Research, 91% of Americans now own a smartphone, a dramatic increase from just 35% in 2011. With this pervasive technology, AI stands poised to revolutionize the way we respond to emergencies. Currently, first responders either lack access to critical information or are inundated with lengthy, convoluted reports that obscure critical information when swift, actionable data is paramount. AI offers the potential to reorganize and streamline this information, providing immediate, clea   s and guidance to first responders on the ground.

A notable innovation in this space is the Haven app developed by RapidSOS, which enables smartphones to automatically transmit a user’s location to 911 with a single touch. This system eliminates the   need   for  individuals  to   convey  vital   information   during the  chaos  of  an  emergency.  Presently,  RapidSOS  processes approximately  25,000  calls  per  month,  representing  a  mere 0.2% of all 911 calls, signifying substantial growth potential. This technology is  poised  to  be  integrated  into  a variety of  devices, from wearables to vehicles to smartphones, ensuring it is available when  emergencies  arise.  Over  21,000  first  responder  agencies and more than 200 technology partners currently connect to 911 through RapidSOS.

‍
As of February 3, 2025, Uber faces 1,562 pending lawsuits in active federal multidistrict litigation, with consumers alleging the company’s failure to protect them from sexual assault or harassment by drivers, per Robert King Law Firm. In light of this alarming figure, AI-powered solutions in the ride-sharing sector could play a pivotal role in facilitating quicker, more informed responses from first responders during such incidents, potentially saving lives and improving safety for passengers and drivers alike.

The global incident and emergency response management market is projected to grow from $137.5 billion in 2024 to
$196.2  billion  in  2030,  at  a  CAGR  of  6.1%, per  MarketsandMarkets.  Startups  such  as Resilitix,  RapidSOS,  and  Omdena,  stand  at the forefront to disrupt this large and growing addressable   market   with   AI   integration. Resilitix, a Texas-based AI startup, offers near- real-time  evacuation  monitoring,  functional criticality  &  vulnerability  assessments,  and seamless integration with existing emergency management frameworks such as NIMS and ICS, among others. RapidSOS, a New York- based  AI  startup,  has  emerged  as  a  go-to AI-powered  emergency  response  solutions provider with its proven benefits to Grubhub, SiriusXM Connect, and Uber, among others. Omdena,  a  California-based  AI  startup,  is trusted by over 300 organization cross the globe, affirming its credibility.