R&D is more important than traction: How Kazakhstan can win in the era of AI startups

In this article, venture investor and author of The Tech, Bakht Niyazov, explains why it is more important for Kazakh AI startups to invest in R&D rather than just focusing on rapid traction. In a previous article, he shared his observations on delivery problems and startups that could emerge from this context, demonstrating the link between real business challenges and innovation.

Bakht Niyazov, city — Almaty, venture investor, Instagram

The industry we refer to as Central Asian venture capital operates by established rules. And that is not a bad thing. Ultimately, money goes to those who already have sales, recurring payments, and a clear LTV/CAC ratio.

However, the artificial intelligence revolution, by definition, implies a paradigm shift not only for startup founders but also for investors; otherwise, one might not win at all, or win less than possible. I am interested in how we ourselves must change to avoid missing out on the strongest teams and the largest future markets. These thoughts apply to the future of Most Ventures as well as other industry players.

Why old approaches don’t work

The problem is that AI startups are DeepTech. And we are not yet accustomed to digging into this. In Kazakhstan, there is no class of investors possessing stable competence in this area. It is impossible not to engage in this field, but there is currently little experience or specific approaches.

The path for AI startups is always longer and more complex. I personally have experience as a co-founder in such a project. First, there is the idea, then long and expensive research, and only then pilots and the final product. Only after that do discussions about scaling begin. At the pre-seed stage, you might have a world-class team, a unique dataset, or something similar, some patents, or a breakthrough model architecture. But such a startup has no revenue, contracts, or dozens of clients.

In short, it is absolutely clear to me that in such projects, the main asset is R&D, not current traction. If we continue to measure them by the rules of e-commerce or SaaS, we will simply lose them: the best will go where they understand the value of a long development cycle.

An example that made me think

Higgsfield AI is our potential organic unicorn project; there is a good chance this will happen. This is a startup in which Murat Abdrakhmanov invested as an angel investor even before the product appeared. He believed in the team and in the fact that our programmers had every chance to move at a global level. The guys have the right DeepTech approach, namely, a focus on the team, research, technological hypothesis, and the future market. Murat recognized this and invested at a very early stage. This is what the best global funds and accelerators are doing now.

And now the question is — how many such examples do we have? And are we ready to create them, if not on a mass scale, then at least as repeatable cases?

Where communication breaks down

The problem of communication between investors and AI startup founders is key. I occasionally help startups for free, just for the sake of the idea. This helps to see the picture from the inside, as startuppers perceive it. I recently met with some cool guys who are building the AVM.AI startup. I was just writing an article about the problem of food delivery services, and this team had already started creating an AI-based platform that solves not only the delivery problem itself but is also aimed at transforming restaurants into AI businesses. I proceed from the currently popular hypothesis that all businesses that transform into AI companies will survive, while others will not.

Three founders, all with relevant higher education, one even holds a PhD degree. All chances for success. Moreover, the guys already have sales and generally decent traction. But the team was at a crossroads. They are required to show traction, forcing them to work with small restaurants in pursuit of MRR. On the other hand, they need to work with a large industry player to leisurely build the product of the future. In the end, I am glad that the AVM.AI founders chose the R&D path. And it is good that among large restaurateurs, there were those willing to help.

As a result, for the startup, it is now important to find understanding among investors that they need at least $500,000 for R&D, which automatically means that the valuation cannot be lower than $5 million. Naturally, investors often try to lower the valuation. But those who understand the industry realize that this contradicts their own interests.

Summarizing this case, I would like to highlight a separate layer of problems for startup founders and for investors.

For founders, the problem is that investors demand metrics that cannot be shown at the R&D stage. It is good that in the case of this startup, there is still some traction. Sometimes inexperienced investors may set inherently incorrect tasks, for example: “connect 10 more clients and come back.” This may yield nothing but a loss of time and resources. And, perhaps, the main problem is the lack of infrastructure in Kazakhstan and Central Asia for cheap and fast research. Large amounts of money are needed.

For investors, the key problem is the lack of expertise. We currently do not have top technical analysts within the country who can evaluate the technical aspects of an innovative product for which there is no analogue yet. There are extremely few of them in the country. Consequently, there is a risk of investing in a technology that ultimately will not reach the market. Fund managers have responsibility toward LPs: they need quick returns, but here there is constant uncertainty.

As a result, of course, many of us would like to repeat Murat Abdrakhmanov’s experience with Higgsfield AI, but not everyone has such an opportunity in terms of expertise.

Why R&D is more important than traction

In the AI economy, the winner is not the one who quickly made an MVP and started selling, but the one who developed a unique technology with a strong team, gained access to critical data, and created barriers for competitors at the architecture and IP levels.

Regarding my own startup, Gen2B, I understand that this may take years. But it is exactly such projects that then explode the market. We all remember that DeepMind, Anthropic, and OpenAI raised hundreds of millions of dollars without a commercial product. And we must prepare ourselves to be maximally competitive specifically in technical expertise.

How to solve the problem and what the new interaction model will look like

I want the first R&D hub to appear within the structure of our Most-holding. I see this as a place where a founder gets access not only and not so much to GPUs or data infrastructure. The participation of titled scientists, experts, and mentors is important. It is important to conduct pilots and allow for the possibility of failure. This is not an accelerator in the classic sense, as we currently have within the Most structure, but a joint laboratory of founders, investors, and technical analysts.

This will help not only startups but also investors. We need investors of a new type — those who are ready to work with the pre-revenue stage, relying on team quality, technology maturity, and market scale.

It is important for us to develop our own AI expertise through such a laboratory so that we do not fall behind founders, and in some aspects, even surpass them. This is primarily about vision.

We talk a lot about the lack of activity in the corporate sector. But within the framework of the laboratory, corporate pilots will be important to us. Corporations should set tasks and provide opportunities for testing. This is a win-win. The startup tests the technology, and the corporation receives innovation without costs for its own R&D. We must work closely with the corporate segment. It is not enough that currently only one Freedom Holding has all the activity.

Quasi-state support will also be important. Subsidies for R&D infrastructure, government orders for AI projects in priority industries: ecology, logistics, medicine, security. Personally, I am optimistic about the new strategy of the Funds Fund “Alem.” I believe the fund can do a lot in this direction by interacting with the private sector.

What Kazakhstan can do

I think there will be a huge boost when we create our own R&D hub for AI startups in MOST. And this should not be just another platform for communication, but an entire environment where:

— founders can quickly test hypotheses and build prototypes
— corporations test technologies before implementation
— investors and analysts learn to evaluate deeptech at early stages
— we grow a new generation of investment analysts who understand AI more deeply than founders.

This is a huge challenge and a huge upside. Moreover, we remain regional leaders and clearly realize that our startups are more than competitive in the world.

If we understand this today, we will not just earn money; we will create an environment where Kazakh talents will develop themselves here and then conquer the whole world.

R&D is our chance for leadership in the future economy.