Fundraising is broken. The user is gone.
An internal note to our portfolio on why 2026 won’t reward “good companies” – only post-labor ones.
Earlier this year, we shared an email with the portfolio. It sparked many valuable conversations, hence we are sharing it here.
---
From: Team NAP
To: NAP Portfolio
Subject: Fundraising in 2026
Date: Tue, 6 Jan 2026
Dear Remi,
Starting off 2026, we wanted to share some thoughts on fundraising this year. We recognize that many of you will be seeking capital, and we believe that several significant, evolving dynamics will be highly relevant to your success.
The shift
2025 marked a fracture in the fundraising landscape. Before then, we operated under a set of agreed-upon laws of physics. We had T2D3, we had specific burn multiples, and we had the magical one million in ARR that signaled readiness for a Series A.
In 2025, the atmosphere changed. It wasn’t just the outlier AI companies hitting tens of millions in revenue seemingly overnight. It was a visceral feeling that the benchmark for growth hadn’t just moved, but that the game itself had been rewritten. The old rules of velocity no longer applied, and something else was at play.
The anomaly
Throughout 2025, we have observed a significant shift in the market that challenges traditional assumptions about fundraising. Companies that, by every metric (10x YoY growth, low burn multiples, excellent NRR), should have been slam dunks are finding it much tougher than expected to attract capital.
They operate in popular sectors, leveraging generative AI, ticking every single box that would have guaranteed a bidding war in the past.
Yet, despite these perfect profiles, many funds known for aggressively backing high-growth companies, hesitated.
The question becomes: what is happening? Why are companies with flawless metrics unable to attract the capital that previously would have flooded in?
The post-labor thesis
We believe the core of this hesitation is a single, existential question: how will this company be successful in a world where human labor is largely gone?
If you position yourself as “the amazing platform for frontline teams,” what happens when there are no more frontline teams? What happens if you sell seats to “sustainability teams,” but those teams are reduced by a factor of ten due to AI efficiency improvements?
Startups take six to ten years to exit. With the current trajectory of AI improvement, most startups founded today will not exit before AGI arrives. While AGI will not distribute to 100% of customers immediately, especially in enterprise settings where rollouts take years, it poses an existential risk to any company not built for that reality.
Round sizes clearly show this pattern. Capital allocation now follows a binary separation. Companies believed to profit from this future raise massive rounds at valuations completely detached from operational reality. Meanwhile, companies that are operationally sound, the “good companies” by yesterday’s standards, struggle to raise a dime.
The invisible user
If you are looking to raise money in 2026, you must nail the positioning for a post-human labor world. To thrive, assume your human buyer will be gone in the next couple of years.
The legal entity you sell to will remain, but the user group operating your product will not be human. It will likely be less than 50% human within two years. Ultimately, the operational use of your product will be through other systems and agents interacting with your software.
This poses a host of uncomfortable challenges. How will you charge for your product? Does your product even need an interface, or is it an API for other agents? How do potential buyers discover your product when the buyer is a script? How do you onboard a non-human customer?
Your strategy needs to weave the path from today to this future.
Equally important is demonstrating the company’s execution capability against that strategy. Investors will prioritize seeing concrete, in-production AI use cases (ideally agentic or nearly so), along with strong AI talent within the team.
The asymmetry of opportunity
Collapsing labor costs and a widely automated landscape does not only come with challenges, but also provide massive opportunities. Here are three examples:
As discovery and purchasing decisions shift to programmatic and self-serve channels (supported by hands-off/agentic onboarding flows), companies can uncouple revenue growth from headcount growth. This is significantly cheaper than employing a fully-fledged sales force.By building software that replaces labor rather than aiding it, total addressable markets become orders of magnitude bigger. You are capturing the value of the salary, not just the software budget.
Agentic coding has become a force to be reckoned with. Until recently, companies were constrained by engineering resources. Agentic coding tools now allow a single team to build ten times more product, enabling them to capture adjacent opportunities much faster.
Speed is of the essence. For every YC batch, dozens of startups are pitching the same idea. Chances are there is at least one “cracked” team somewhere 996ing the same thing as you. Companies that get this right will win. Companies that do not will have no future. It is both the opportunity of a lifetime and an existential threat.
The great divergence
We are not sure if all VCs are consciously thinking about investing in post-labor companies. Some certainly do; returning a billion-dollar early-stage fund for managers like Accel, Index or Founders Fund requires massive outcomes. But others likely follow the herd, mimicking the big brands.
In any case, there is a clear separation between the haves and have-nots. We expect this trend to continue to an extreme stage, where some companies command incredible valuations while others are unable to raise any money at all.
The three imperatives
Beyond addressing the post-labor thesis, three elements have shifted from being important to being absolute prerequisites.
First, great storytelling has become absolutely essential (and leverage if done right). Founders need to clearly convey the vision in a way that gets investors excited about the future state. Many founders get stuck in the weeds of the “now” and miss the opportunity to tell the big story.
Second, nailing your brand and visual identity is non-negotiable. In a world where any feature or product can be replicated within hours, brand becomes a moat. Investors see brand DNA (and related strong customer relationships) as a critical asset that cannot be forked or copied. Do not half-ass this.
Third, you must optimize for velocity. Build systems that allow the company to move at maximum speed. Hire talent that leverages the latest technologies. Particularly in product engineering, failing to use agentic coding tools is no longer a preference; it is a red flag.
Closing thoughts
This shift will not affect all sectors identically. A company building Infra or selling into Defense or Energy may not face the immediate, existential threat of a disappearing user base in the same way a B2B SaaS platform does. Similarly, the dynamics of Consumer AI, where the product is the agent itself, are radically different. However, no sector is immune to the underlying seismic change. Even if the post-labor negative effects don’t apply, every company must recognize opportunities such as the acceleration of product development, the near-zero customer acquisition costs, and the explosion of TAM driven by capturing labor value. The mandate is universal: ride the resulting wave of opportunity and profit from the new efficiency, or be outpaced by those who do.
We hope this is helpful. Let’s go and crush it!
Your team at NAP 💜
PS: Here’s a highly recommended related post on Context graphs and why they are AI’s trillion-dollar opportunity.
Interesting framing. It connects directly to Gupta/Garg's "context graphs" piece you linked - the next platforms won't be systems of record for data, but for decisions: the reasoning and precedents that currently live in people's heads.
We're building exactly this at Xemantic. Our open-source Golem XIV (https://github.com/xemantic/golem-xiv) uses Neo4j knowledge graphs as persistent decision memory, with agents expressing reasoning through executable Golem Script (internal programming language expressing cognitive process) - every decision trace is replayable by construction, and the architecture is LLM-independent. Legal tech is our first domain. German federal law is pure precedent and exception logic, exactly the signal Gupta identifies. Autonomous scientific research comes as a next frontier.