Sunburst Interactive memo

At 15, I wanted to build a game about the Biafran war. At 16, I was already a game developer at one of the top studios in Nigeria.

Then my path turned to payments. At 19, I founded Lazerpay. I built it for the right reason. I cared about the problem, and I wanted to fix it. We shut Lazerpay down and Paystack acqui-hired the team.

After Lazerpay, something changed. I wanted to win, and I wanted it fast. I went after whatever looked like the quickest result, I felt I needed to prove something. I built Ultramarkets, and it sold too. But somewhere in that run, I stopped building for the reason I started.

So I stepped back. I took a break from companies to find out what I actually want to build.

The answer was the game I wanted to make at 15.

Rising Sun is that game. Sunburst is the company I am building to make it, and the stories that come after it.

Preface

Today, a AAA game takes 5 to 10 years to make. It needs hundreds of people and a budget that only a few companies can pay.

This is why the stories in games come from so few places. Africa has 54 countries, thousands of languages and centuries of history. No African studio has made a AAA game for the world.

I do not think this lasts. Within the next decade, I believe a small team will make a AAA game in a fraction of the time it takes today. The tools to do that do not exist yet. We are building them.

Specifically, Sunburst Interactive is an interactive studio and a research lab. The Studio makes hyper-realistic AAA games that tell African stories. Our first title is Rising Sun, set in the Nigerian Civil War. We make it by hand, with no generative AI. The Lab records how our experts make it, and trains models on that work. Each game we make teaches the models. The models make the next game faster.

Software creation is now open to almost anyone. Game creation is next. The company that owns expert production data will own that change.

Intro

AAA games have always been made by a small number of studios in a small number of countries. The US, Japan, Western Europe and, more recently, China. The stories follow the studios.

This model is now at its limit. Production cost and time go up with each generation. Insomniac's own files showed that Spider-Man 2 cost about $300 million, almost three times the first game (Kotaku). In the same files, the studio asked if players can even see that extra investment.

Two things are true at the same time:

1. Players want new worlds. When a studio from a new place tells its own story at AAA quality, the world buys it.
2. The cost to make that game is so high that almost no new studio can try.

Black Myth: Wukong shows the first point. Game Science, a Chinese studio, made a AAA game from the Journey to the West myth. It sold more than 10 million copies in three days (Gulf News). It has now passed 30 million copies, and more than half of sales came from outside China (Digital Citizen).

AI shows a way past the second point. But AI can only make AAA work if it learns from AAA experts. Today it does not.

This memo explains the problem, the opportunity, and how Sunburst plans to win both.

Background

For simplicity, I split the problem into two parts: stories and production.

Stories are what a game says and who it is about. Production is the work that turns a story into a playable world. Sunburst has one unit for each. The Studio owns stories. The Lab owns production.

It is also important to define the parts of AAA production. Each part has different experts, tools and data. Each part is a separate problem for AI.

Characters

The people of the game. Concept, sculpt, retopology, textures, hair, rigging and facial setup. A main character in a AAA game can take a team of artists months. This is where African faces, skin tones and hair are most often wrong in today's tools.

Environments

The world the player moves through. Landscapes, buildings, props and materials. For Rising Sun, this means 1960s Nigeria: towns, roads, bush, villages and military sites, all based on historical references.

Animation

How characters and objects move. Motion capture, cleanup, hand-keyed animation and cinematics.

Lighting, audio and assembly

Lighting, sound, music, level building and the final assembly in the engine. These steps join all the other work into one experience.

The Lab starts with characters and environments. They take the most expert time, and they are where our African data gives the largest advantage.

The problems

With the two parts defined, these are the problems in each.

Production cost grows faster than revenue

Each console generation asks for more detail: more polygons, more animation, more cinematics. The work is still done by hand, asset by asset. So cost goes up in a straight line with ambition.

Insomniac's internal slides set a target to keep future AAA franchise games at $350 million or less (Kotaku). That is the ceiling a major studio thinks is sustainable. No new studio can enter at that price.

AI tools learn from the wrong data

Today's generative 3D tools learn mostly from final assets. A final asset shows the result. It does not show the decisions: why an artist moved an edge loop, how they fixed a deformation at the shoulder, which reference they used for a fabric.

This is why most generated assets need heavy cleanup before a AAA team can use them. The model never saw how an expert works. It only saw what experts left behind.

Expert knowledge is never recorded

In every studio, the most valuable knowledge sits in the heads of senior artists. It moves from person to person, over the shoulder. When an artist leaves, the knowledge leaves. No studio records it in a form a model can learn from.

Africa is missing from the data

General models are weak where their training data is weak. African faces, skin tones under different light, natural hair, textiles and architecture are badly represented. The output looks generic or wrong. A studio that wants to tell an African story at AAA quality cannot rely on these tools.

African stories have no AAA home

Africa has talented artists and developers. Many work remotely for studios abroad, on other people's stories. There is no AAA studio on the continent to bring that talent together around African stories, at the standard global players expect.

The opportunities

The global games market made $201.6 billion in 2025, up 9.1%. Premium PC game spending grew 25.3% to $14.1 billion (GamesBeat). Players pay for premium, story-driven games. Wukong showed they pay for new cultures too.

But the larger opportunity is the cost of production itself. Every AAA studio spends most of its budget on people doing work by hand. A tool that cuts that time is valuable to every one of them.

The first AAA game out of Africa

Rising Sun is set in the Nigerian Civil War. The player sees the war through personal stories, real events and real places. We make it by hand, from scratch, with no generative AI. A playable demo exists.

The first AAA game from a new region sets the standard for that region. Game Science is now the reference for Chinese AAA. Sunburst can be that reference for Africa. Rising Sun is the first title in a franchise of African stories.

Expert process data

Because we make Rising Sun by hand, we can record how it is made. We capture the steps our experts take in Blender and Unreal Engine 5: each edit, each tool, each fix. Every contributor contract gives Sunburst the right to train on this work.

This is data that no scraper can collect. It does not exist on the internet. It only exists inside a studio that decides to record it.

Nkata: the tool is the distribution

To record process data, the work has to happen somewhere. That somewhere is Nkata.

Nkata is where a game team makes the game together, inside Blender and Unreal. Every save in Blender shows up in Unreal in seconds. Every part of an asset has an owner. Every review happens on the model, in the engine. One line: Figma for game development, with real files you keep.

We built it because the Rising Sun team needed it. Perforce is the industry standard, and its setup stops a small studio from adding a second artist. Our artists were sharing work on Discord. No tool in this space is one an artist wants to open.

Nkata records process as a side effect of normal work. Every saved version. Every approve and send back. Every note pinned to a point on the mesh. Nothing extra for the artist to do.

Two plans, same product. Free studios get full Nkata, and their process data trains Nok. Premium studios pay per seat, and their data is never used for training. The studio chooses, every member can see the setting, and off means off.

This is how the dataset grows past Sunburst. Every free studio makes Nok better. A better Nok makes Nkata more valuable. The tool is the distribution. The data is the moat.

Models for each production step

The Lab trains models on that data, one production step at a time. We start with characters and environments. Then animation, lighting and level building.

The goal is not a model that makes a whole game from one prompt. The goal is a set of models that each do one step at expert level, inside the tools artists already use.

The African wedge

African faces, skin tones, hair, textiles and architecture are where general models fail most. They are also where our data is strongest. This gives the Lab a clear first market that large players do not serve.

The flywheel

1. A studio makes a game in Nkata.
2. Nkata captures how the experts make it: versions, reviews, corrections.
3. The Lab trains Nok on that data.
4. Nok comes back into Nkata as a tool in the same panel.
5. The next game is faster to make. More studios join. More data. Nok improves.

The loop starts with Sunburst. Rising Sun is the first game made in Nkata. But the loop is not limited to us. It runs for every studio on the free plan.

There is a reference case in film. InterPositive, the AI filmmaking company that Ben Affleck founded, trained models on footage that filmmakers already shot. It had about 12 staff. Netflix acquired it in March 2026 (NPR). InterPositive works in postproduction. Sunburst goes further: our models make production assets.

Existing players

Global AAA studios

They have budget, teams and distribution. Some have set games in Africa. But they tell these stories from the outside, and their cost structure is the problem we describe above. They are also possible future customers of the Lab.

Engine and character tools

Epic's MetaHuman makes realistic digital humans inside Unreal Engine. It is strong, widely used and close to the artist. It is built from Epic's own capture data, not from the process of a studio making a full game. It is the closest tool to our first Lab target, and the one we must be better than for African characters.

Generative 3D startups

Many well-funded companies turn text or images into 3D assets. Their output is good enough for prototypes and small games. Most do not yet reach AAA standard, because they train on final assets, not on expert process.

African game studios

Studios across the continent make mobile, casual and indie games, and some make larger titles. They prove the talent exists. None has yet shipped a AAA game for the global market.

Studio collaboration tools

Perforce is the standard for version control in games. Anchorpoint and Diversion are easier, but keep the same shape: file trees, commits, branches. Built by engineers for engineers. None of them has presence, checks before publish, review in the engine, or comments on the asset itself. None of them captures process data. This is the gap Nkata fills.

Game Science

Not a competitor, but the closest precedent. A studio that made a AAA game from its own culture, at a fraction of the cost of a Western AAA game, and sold it to the world. Reports put its development cost at around $56 million (ARY News, citing the Financial Times).

Distribution

The Studio and the Lab reach their markets in different ways.

Trailer first

Wukong's first trailer came out in 2020, four years before the game. It got global attention and brought strategic investment from Tencent (Genuine Impact). The trailer drew 10 million YouTube views outside China (ARY News).

We follow the same path. Our pre-seed round funds a public Rising Sun trailer. The trailer builds the audience, wishlists and press before the full game exists. It also proves our quality to the investors for the next round.

Culture as a channel

African music, film and fashion already travel the world. The audience for African stories is global, and it is already in place. Rising Sun speaks first to Nigerians and the diaspora, who know this history from their families. Then to every player who wants a new world.

Premium PC and console

Rising Sun is a premium cinematic action strategy game. A limited free demo lets players try it. Then they pay one time for the full game. Premium spending grows fastest on PC.

The Lab: our studio first, then the tools artists use

The first customer of the Lab is the Sunburst Studio. Every model is tested on our own production before anyone else sees it.

After that, the Lab reaches other studios through Nkata. The free plan is the offer: a better tool than Perforce, at no cost, for studios that let their process data train Nok. Nok-0, the first model, ships as one button in the Nkata Blender panel. The tool carries the model to every studio that uses it.

Risks

Six risks hold real merit.

Execution

AAA quality is hard. Wukong cost a reported $56 million and took a team of veterans years. Our pre-seed is $3 million. It does not make a full game, and we do not claim it does. It makes a trailer, starts production and starts the Lab. The second raise funds the full build. Each raise must earn the next one.

Data volume

One hand-made game gives a small dataset compared to what large models train on. Our answer is Nkata's free plan. Every studio that uses Nkata for free adds its process data to Nok. If no outside studio adopts Nkata, the dataset stays small and Nok stays a tool for our own studio. The second phase of the roadmap has one gate: one outside studio uses Nkata daily. We do not move past it until that is true.

Large players build it first

Epic, the large AI labs or a well-funded 3D startup may build expert-level asset tools before us. They have more money. They do not have a studio recording AAA process data on African content. Our data and our studio are the defense. Speed of capture matters.

Artist trust

Many artists and players distrust generative AI in games. Much of that distrust comes from models trained on work without consent. Our model is the opposite: artists are paid, contracts state the training rights, and Rising Sun itself is made by hand. We must keep that promise in public.

The subject

The Nigerian Civil War is still a sensitive subject in Nigeria. A game about it will draw attention and criticism. We show the war from more than one perspective, with historical advisers, and with care for the people who lived it. Handled well, this is the reason the game matters. Handled badly, it harms the company.

Focus

Three products in development at pre-seed looks like three companies. It is one loop, built in order. The pre-seed funds Rising Sun only. Nkata exists because the Rising Sun team needs it today, and it is built by the same small team. Nok starts with the first ML hires and ships inside Nkata. Each phase has a gate, and we do not start the next one until the gate is passed.

Why now

• AI changed software first. Coding models made one engineer do the work of many. They got there by learning from expert work. Games are next, and the same rule applies.
• Engines made photoreal possible for small teams. Unreal Engine 5 gives a small studio tools that only the largest studios had ten years ago.
• The world proved it wants new cultures. Wukong passed 30 million copies, mostly outside its home market.
• Studios need to cut cost. AAA budgets are at the limit of what even the largest publishers think is sustainable.
• No one records expert process yet. The first studio that does it builds a dataset that others cannot copy later.

Why us

• We are the people whose story this is. Rising Sun is our history, told from the inside.
• We already started. A playable demo exists, made by hand.
• We build the capture tools ourselves. Nkata sits inside Blender and Unreal Engine 5, where the work happens.
• I started in games. At 16, I was a game developer at one of the top game studios in Nigeria.
• I am a founder with two exits. I founded Lazerpay at 19. It served more than 3,000 merchants and was acqui-hired by Paystack. My second company, Ultramarkets, was also acquired.