The Chip That Built the World (And Why Africa Needs to Own a Piece of the Next One)
By Oluwaseyi Ayodeji Published on oluwaseyiayodeji.com | Sovereign Stack Newsletter
“You cannot negotiate sovereignty from a position of dependency. You must first own something the world needs.”
SERIES BRIDGE
In Issue 02, we examined Talent - the human infrastructure that makes AI possible. In Issue 03, we looked at Compute - the physical machines that run AI workloads. Both arguments led to the same uncomfortable question: What happens when the chips that power those machines are manufactured entirely elsewhere, from materials largely extracted from African soil, with none of the economic value captured on the continent?
Today, in Issue 04, we go upstream. We go to the chip itself.
THE MOMENT I UNDERSTOOD WHAT A CHIP ACTUALLY IS
In 2013, I walked into a semiconductor fabrication facility in Hillsboro, Oregon - one of Intel's flagship fabs on the sprawling campus that the company had been building since the 1970s. I had just completed my Master's degree in Materials Science and Engineering, and I had accepted a process engineering role. I thought I understood what I was walking into.
I did not.
I was assigned to the CDO team - Carbon Doped Oxide - working in the back end of the fab's manufacturing process. CDO is a dielectric material deposited using Plasma Enhanced Chemical Vapor Deposition (PECVD): a technique that uses plasma energy to drive a chemical reaction that coats the wafer surface with ultra-thin, precisely controlled layers of material. My team's work was part of building the interconnect layers that allow billions of transistors to communicate with each other inside a chip.
One step. Of hundreds.
That realization changed how I understood everything. I spent the next five years at Intel - working across multiple fabs: D1C, D1X, D1D - not just executing my role, but pulling on every thread I could find. I read about lithography. I talked to the equipment vendors. I traced the supply chain backward to the raw materials and forward to the customers. I wanted to understand the full system, not just my seat in it.
What I found was one of the most complex, geographically concentrated, and strategically consequential manufacturing ecosystems in human history.
And it is the ecosystem that determines who builds AI - and who merely uses it.
WHAT A CHIP ACTUALLY IS (FOR EVERYONE WHO WASN'T IN THE FAB)
Before we can talk about strategy, we need shared language. Here is the journey of a chip, from sand to supercomputer, stripped to its essentials.
Step 1: Silicon from Sand It starts, improbably, with silica sand - one of the most abundant materials on earth. Through a series of energy-intensive purification processes, that sand is refined into polysilicon with a purity level of 99.9999999% - nine nines of purity. That number matters. A single atom of contamination in the wrong place can kill a transistor. The purified silicon is melted and grown into cylindrical ingots using the Czochralski process, then sliced into ultra-thin wafers - the blank canvases on which chips are painted.
Step 2: Photolithography - Painting with Light This is where the magic and the geopolitics collide. Photolithography is the process of using light to transfer a pattern - essentially a circuit blueprint - onto the wafer's surface. The wafer is coated with a light-sensitive material called a photoresist. A mask carrying the circuit pattern is aligned over it. Light is passed through the mask, hardening or softening the resist in the exact pattern of the circuit. The unhardened resist is washed away, leaving the circuit pattern etched in material.
Modern chips require dozens of these lithography steps, each one stacking another layer of circuitry. The most advanced chips today are built using Extreme Ultraviolet (EUV) light - a wavelength so short it must be generated in near-vacuum conditions and cannot be projected through conventional optics. There is exactly one company in the world that manufactures EUV lithography machines: ASML, a Dutch company based in Eindhoven.
Step 3: Deposition, Etch, and Planarization Between each lithography step, engineers deposit new materials onto the wafer (as I did with PECVD), etch away unwanted material using plasma (dry etch) or chemical solutions (wet etch), and then use a process called CMP - Chemical Mechanical Planarization - to polish the wafer surface perfectly flat so the next layer can be applied uniformly. A deviation of a few nanometers in any of these steps cascades into defects that render entire wafers unusable.
Step 4: Testing, Dicing, and Packaging Once all layers are complete, the wafer is tested electrically to identify functional chips. It is then cut - diced - into individual dies. Each functional die is mounted, wired, and encased in a protective package that allows it to interface with the outside world. This packaging step, long considered a commodity process, has become one of the most contested and innovative areas of the entire industry as chipmakers race to integrate multiple dies into single packages for maximum performance.
What I have just described - compressed into five paragraphs - is the work of tens of thousands of engineers, hundreds of specialized equipment vendors, and supply chains spanning a dozen countries. All of it, to make something smaller than your thumbnail that contains more transistors than there are stars in the Milky Way galaxy.
THE ARCHITECTURE OF LEVERAGE: WHERE POWER ACTUALLY SITS
Here is the uncomfortable truth about semiconductor manufacturing: it is not a level playing field. It is a cascade of near-monopolies, each one a chokepoint that whoever controls it can - and does - use as a weapon.
ASML and the Lithography Lock The most dramatic example of concentrated leverage in the entire technology economy is ASML. This single Dutch company holds 100% of the market for EUV lithography machines - the equipment required to manufacture chips at 5nm, 3nm, and 2nm process nodes, which is where all AI accelerators and advanced GPUs are built. There is no second supplier. The technology took more than 30 years and over €10 billion in R&D to develop, and it is protected by over 16,000 active patents.
Each EUV machine costs over $400 million. Each one contains more than 100,000 parts sourced from thousands of suppliers across dozens of countries. ASML's backlog at the end of 2025 sat at €38.8 billion - meaning demand is so far ahead of production capacity that customers wait years for delivery. In 2025, ASML reported net sales of €32.6 billion, a 15% year-over-year increase.
When the United States government wants to prevent China from advancing its semiconductor manufacturing capabilities, the primary tool it reaches for is restricting ASML's ability to export machines to Chinese customers. That is the measure of ASML's strategic centrality. A Dutch company headquartered in a mid-sized city is, effectively, the gatekeeper to advanced AI hardware - a responsibility that ASML's President and CEO Christophe Fouquet has described as one that extends far beyond commerce into the architecture of global security.
TSMC and the Foundry Model When Intel was building my fabs in Hillsboro, it operated under what is called the Integrated Device Manufacturer (IDM) model: Intel designed its chips and Intel built them. This gave Intel enormous control over its manufacturing process, but it also meant that Intel assumed the massive capital cost and complexity of running fabs.
TSMC chose a different path. Taiwan Semiconductor Manufacturing Company would build chips it did not design - it would manufacture designs submitted by external clients. This foundry model turned out to be transformational. It allowed fabless companies - companies with brilliant chip designers but no fabs - to bring products to market without building billion-dollar facilities. And it allowed clients to get custom chips optimized precisely for their own applications, rather than the general-purpose CPUs Intel was mass-producing.
The results have been historic. TSMC today controls approximately 71% of global semiconductor manufacturing market share, under the leadership of Chairman and CEO C.C. Wei. Its most important client, NVIDIA - led by founder Jensen Huang - submits its GPU designs to TSMC for fabrication. NVIDIA's market capitalization has climbed to nearly $5 trillion - the highest of any company in the world - built largely on chips it does not manufacture, in a country it does not control.
The Gallium Question Here is where the story takes a turn that should matter deeply to every African policymaker, investor, and entrepreneur reading this.
Silicon has dominated semiconductor manufacturing since the 1960s. It is abundant, well-understood, and highly refined. But silicon has physical limits. As AI workloads scale - NVIDIA's Blackwell GPU chips already consume more than 1,000 watts per unit, requiring radical rethinking of power systems in data centers - silicon-based chips increasingly struggle with heat management and energy efficiency.
Enter gallium. Specifically, Gallium Nitride (GaN), a wide-bandgap semiconductor material with properties that silicon simply cannot match for certain applications. GaN can handle significantly higher electric fields, operate at far greater temperatures, and switch at speeds up to 100 times faster than silicon - all while converting power with efficiency above 98%. GaN-based chips are already being incorporated into NVIDIA's AI factory architecture. GaN is not a future technology. It is here now, and it is scaling.
Here is the critical fact: gallium is not directly mined. It is recovered almost entirely as a byproduct of processing bauxite ore for aluminum production. And the world's bauxite is not evenly distributed.
Guinea alone is the world's largest bauxite producer, with output approaching 200 million tonnes in 2025. Ghana holds an estimated 920 million tonnes of bauxite reserves. Africa as a whole holds approximately 30% of the world's critical mineral reserves, including materials essential to semiconductor manufacturing: gallium (via bauxite), cobalt (DRC supplies over 70% of global production), tantalum, graphite, rare earth elements, and more.
China understood this before anyone else. It currently processes 99% of the world's refined gallium and 85% of global silicon. In December 2024, Beijing imposed export restrictions on gallium, germanium, and antimony to the United States - turning obscure byproduct metals into weapons of geopolitical pressure overnight. The GaN semiconductor device market alone is projected to grow from $3.06 billion in 2024 to $12.47 billion by 2030, a CAGR of approximately 27%.
Africa is sitting on the inputs to the next generation of AI chips. And it is largely watching others process, refine, and profit from them.
THE AFRICA CHIPS DIAGNOSIS: WHERE WE ARE
Let us be precise about the current state, because vague optimism is not a strategy.
What Africa has:
Approximately 30% of global critical mineral reserves, including key semiconductor inputs
The world's largest bauxite deposits (Guinea, Ghana), which are the primary source for gallium extraction
Over 70% of global cobalt (DRC), essential for advanced electronics and batteries
Significant tantalum deposits (DRC, Nigeria), used in capacitors throughout chip packaging
A rapidly growing population of STEM graduates and engineering talent
Countries - Kenya, Rwanda, Ghana, Nigeria - building genuine innovation ecosystems
What Africa does not yet have:
Domestic mineral refining capacity at scale (most extraction still leaves as raw ore)
A semiconductor design ecosystem generating original IP
Chip packaging or testing facilities of significance
The political coordination mechanisms to negotiate collectively at the global table
Reliable industrial-grade power supply at the scale advanced manufacturing requires
The gap between what Africa has and what Africa captures is precisely where the strategic opportunity lives. And critically, that gap is being exploited in real time. China's dominance of gallium refining did not happen by accident - it happened through decades of deliberate industrial policy. Africa's extraction without value capture did not happen by accident either - it happened through a combination of colonial infrastructure design, under-investment in processing capability, and governance frameworks that prioritized volume over value.
None of those are permanent conditions.
AFRICA'S LEVERAGE POINTS: WHERE TO APPLY PRESSURE
The honest version of a chips sovereignty strategy for Africa does not begin with building a leading-edge fab. That is a 20-year conversation requiring hundreds of billions in capital, stable power supply, a mature engineering workforce, and equipment access that is currently controlled by geopolitical rivals. Starting there is not ambition - it is distraction.
The honest version starts with what Africa already controls, and works forward from there.
Leverage Point 1: Minerals as Negotiating Currency, Not Just Export Revenue The DRC's designation of cobalt as a "Strategic Mineral" - enabling stricter extraction and export regulation - is a template. Guinea's revocation of bauxite concessions from foreign operators in 2025, while messy in execution, signals a growing understanding that resource ownership is not the same as resource control. African nations must move from selling raw ore to conditioning access to that ore on technology transfer, local processing requirements, and equity participation in downstream manufacturing.
Botswana's diamond policy - requiring that rough diamonds be cut and polished locally - is decades old and imperfect. But it is directionally correct. The same logic applied to gallium-bearing bauxite would change the equation.
Leverage Point 2: Chip Packaging and Testing - The Entry Point Between the fab and the finished chip sits a critically important step: packaging and testing. Advanced packaging - the process of assembling multiple chip dies into integrated modules - has become one of the most technically demanding and economically valuable parts of the semiconductor supply chain. It is also, relative to front-end fab construction, more accessible. The capital requirements are lower. The technical skills required overlap with existing manufacturing engineering talent pools. Malaysia and Vietnam built significant semiconductor ecosystems starting precisely here. Africa can too.
Leverage Point 3: Chip Design for Africa's Problems ASML's machines are inaccessible to Africa today. The talent and tools for chip design are not. Companies like ARM build chip designs that others license and manufacture. Dr. Lisa Su's transformation of AMD - from near-collapse to a $675 billion AI chip powerhouse competing directly with NVIDIA - is proof that fabless-first design strategy, executed with focus, can reshape an entire industry. Africa has the engineering talent - increasingly - to participate in fabless chip design optimized for African deployment conditions: low-power edge AI for agricultural monitoring, grid-independent sensor networks, climate-resilient IoT. These are not niche applications. They are the foundation of the AI economy that serves African needs.
Leverage Point 4: The Geopolitical Moment China's export restrictions on gallium, germanium, and antimony have created what ODI researchers described as a "fortuitous" moment for African countries to position themselves as alternative sources of semiconductor-critical minerals for Western supply chains desperate for diversification. The US Minerals Security Partnership, the EU Critical Raw Materials Act, and the US CHIPS Act have all created financial and diplomatic infrastructure designed to pull alternative mineral suppliers into trusted supply chain networks. African nations that move with coordination and intention in the next three to five years can capture positioning that will be difficult to dislodge.
THE TRADE-OFF AFRICA MUST RECKON WITH
There is a version of the minerals strategy that fails, and it is worth naming it plainly.
If African governments respond to Western demand for mineral supply chain diversification the same way they responded to Chinese demand - granting concessions, accepting infrastructure loans, exporting raw ore - then the continent will have traded one form of dependency for another. The flag over the mining operation will change. The dynamic will not.
The minerals leverage only translates into chips sovereignty if African nations can negotiate processing and value-addition requirements into the terms of access. That requires political coordination at a continental level that the African Union has historically struggled to deliver. It requires governance frameworks that are transparent enough to attract Western institutional capital without being captured by the political interests that historically exploit resource revenues.
These are real obstacles. They are not insurmountable - but pretending they do not exist produces analysis that sounds good and changes nothing.
SCORING THE CHIPS PILLAR
In the Stack Sovereignty Test framework, the Chips pillar asks: Does this nation or region have meaningful control over the hardware inputs that power its AI economy - from raw materials through design and fabrication to the finished chip?
Africa's current score: 2 / 5
Raw materials: Genuine strength. Africa holds the inputs. ✓
Refining and processing: Near-zero domestic capacity. Critical gap. ✗
Chip design: Emerging, but fragmented. Partial credit. △
Fabrication: Absent. ✗
Packaging and testing: Nascent at best. ✗
A score of 2 is not a verdict of failure. It is a map of where to build. The minerals advantage is real and time-sensitive. The design opportunity is real and growing. Packaging is achievable within a decade with the right policy architecture. Fabrication is the long game - and long games require you to be in the room, accumulating leverage, before the decisive moment arrives.
THE NORTH STAR
When I left Intel in 2018, I carried something more valuable than five years of process engineering experience. I carried a mental model of what a complete industrial ecosystem looks like - how every component, every vendor, every design decision is part of a system that either distributes power or concentrates it.
The global semiconductor industry has concentrated power in fewer hands than almost any other industry on earth. ASML in one Dutch city. TSMC on one island. Gallium refining in one country. This concentration did not happen because it was the only possible design. It happened because of deliberate investment, policy choices, and sustained strategic intent over decades.
Africa's path to chips sovereignty requires the same. Not in one country, not in one company, not in one decade. But with the same clarity of intent: we own the inputs to the chips that power AI. We will not leave that leverage on the table while others build the future with our materials. Nigeria's Minister of Communications, Innovation and Digital Economy, Dr. Bosun Tijani, framed it precisely when he said Africa must move from being a passive recipient of technology to an active architect of its own digital future. Chips sovereignty is where that architecture begins.
The sand beneath African soil was there before Silicon Valley. It will be there after whatever comes next.
The question is who decides what to build with it.
WHAT'S NEXT
In Issue 05, we move to Governance - the policy and institutional architecture without which none of the other pillars can hold. Africa has the minerals, the talent, and the growing compute footprint. But sovereign AI cannot be built on extractive legal frameworks, fragmented regulatory environments, and data governance structures written by others for others. The Governance pillar is where strategic intent either gets encoded into law - or quietly handed away at the negotiating table.
Oluwaseyi Ayodeji is an AI strategist and infrastructure architect who builds end-to-end frameworks that position organizations and nation-states for strategic advantage in the global AI economy. He is the creator of the Stack Sovereignty Test and author of the Sovereign Stack newsletter. He spent five years as a process engineer inside Intel's Hillsboro, Oregon fabrication network - one of the most advanced semiconductor manufacturing campuses in the world.
© 2026 Oluwaseyi Ayodeji | oluwaseyiayodeji.com
