The next big headache in technology isn’t computing power or artificial intelligence. It’s memory. As AI data centres, autonomous vehicles and humanoid robots develop an insatiable appetite for chips, Elon Musk is contemplating a staggering solution: Terafab, his own semiconductor manufacturing venture. Musk says his companies’ future chip requirements could exceed the world’s entire current production. Ponder that for a moment. Without Terafab, if he tried to secure that supply from existing manufacturers, nothing would be left for the rest of us. It is a frightening prospect – and a timely warning for Kenya, which is still figuring out its place in the global digital economy.
Elon Musk wants to colonise Mars, populate the world with humanoid robots, turn cars into autonomous machines and, eventually, move artificial intelligence infrastructure into space.
It is an extraordinary vision, even by Musk’s standards. But there is a rather earthly problem threatening to slow it down: Computer memory. You see, all these sophisticated machines need enormous amounts of memory to function. They need somewhere to store the information they are working with and, more importantly, the ability to retrieve it almost instantaneously. Without sufficient memory, even the world’s most powerful processors become expensive, underutilised pieces of silicon.
And Musk is beginning to worry that the world simply won’t produce enough chips to feed his ambitions.
So serious is the challenge that he is pursuing an ambitious semiconductor manufacturing venture called Terafab, bringing together the interests of Tesla, SpaceX and xAI. The objective is to develop the chip manufacturing capacity needed to power his increasingly ambitious technology empire.
In March 2026, announcing the project in Austin, Texas, Musk said his companies’ future requirements would eventually exceed the entire world’s current chip production.
“We either build the Terafab or we don’t have the chips,” he declared.
According to Reuters’ account of the Terafab announcement, the plan involves two advanced chip factories: One to produce processors for Tesla vehicles and Optimus humanoid robots, and another to support AI data centres in space.
The ambition is enormous. By October 2026, Musk was targeting volume production by the third quarter of 2028, although he acknowledged that the timetable was extremely aggressive. The factories are intended to manufacture advanced logic and memory chips, as well as handle packaging and testing.
But here is a nugget worth pondering.
If Musk’s ambitions materialise on the scale he envisages, and he has to depend entirely on existing manufacturers to supply his chips, what happens to everybody else?
An estimate by John Lorenz, director of computing and memory at semiconductor research firm Yole Group, quoted by EE Times, suggests that Terafab’s eventual memory requirements could reach 10 million wafers a month – five times current global DRAM production.
That is an extraordinary projection, not an established order or guaranteed requirement. But it illustrates the scale of the potential demand problem.
Imagine a world in which one technology empire could theoretically require several times the memory the entire industry currently produces.
What would be left for the rest of us?
Reuters’ October 2026 update on Terafab provides a glimpse of the scale of the undertaking. The EE Times analysis of Terafab’s memory requirements makes the implications even more striking.
And this is where the story gets interesting for Kenya.
What appears to be a problem for Silicon Valley could become one of the most consequential economic and industrial developments of the next decade, particularly for countries still trying to establish their place in the global digital economy.
To understand why, we need to appreciate something often overlooked in our technology conversations: The importance of computer memory.
The new oil may be silicon
For years, the technology race has been about processing power. Who has the fastest computer? Who has the most powerful graphics processing units (GPUs)? Who can build the largest artificial intelligence model?
But powerful processors need equally capable memory systems to function efficiently. Artificial intelligence has made this relationship particularly important. Training and running sophisticated AI models requires enormous amounts of data to move rapidly between processors and memory. This has created extraordinary demand for High Bandwidth Memory (HBM), a specialised form of technology designed to feed powerful AI processors with information at very high speeds.
The problem is that manufacturing these specialised memory chips is complicated, expensive and capacity-constrained.
The global memory industry is dominated by three major manufacturers: Samsung, SK Hynix and Micron. Their production decisions increasingly influence the availability and pricing of memory across the wider technology industry.
The CSIS analysis of the AI-driven memory crunch explains why this is becoming a strategic issue rather than simply another cyclical shortage in the semiconductor market.
And the consequences could extend well beyond AI companies.
When robots start competing with laptops for memory
Consider Musk’s ambitions.
Every humanoid robot needs memory to process its surroundings, interpret instructions, recognise objects and make decisions. Autonomous vehicles need it to interpret camera feeds, sensor information and road conditions. AI data centres need enormous quantities of it to run increasingly sophisticated models.
Now multiply those requirements by millions of robots, vehicles and AI-powered systems.
We are no longer talking about a handful of technology companies buying more computer memory. We are talking about an entirely new generation of machines competing for the same semiconductor manufacturing capacity.
The pressure is already being felt across the industry. AI companies are competing for specialised memory, while manufacturers are allocating more resources to higher-margin products. This can squeeze the supply of conventional Dynamic Random Access Memory (DRAM), the technology behind most conventional computer RAM, used in laptops, smartphones, gaming consoles and other everyday electronics.
For ordinary consumers, this could mean more expensive computers, phones and electronic equipment. For businesses, it could mean higher infrastructure costs. For countries dependent on imported technology, it could mean increased exposure to global supply disruptions.
And there is another complication: Building semiconductor factories takes years, requires enormous capital and depends on specialised equipment, technical skills and reliable infrastructure.
This is not an industry where you can simply increase production overnight because demand has suddenly surged.
Musk’s Terafab ambitions are therefore revealing. The world’s wealthiest technology entrepreneurs are beginning to recognise that controlling access to semiconductor capacity could be just as important as developing the next breakthrough AI model.
Kenya’s Silicon Savannah moment
Kenya should pay attention because we are approaching this technological revolution from an interesting, but potentially vulnerable, position.
For years, we have proudly promoted ourselves as the Silicon Savannah. We have built a reputation around mobile money, fintech, digital banking, software development and a growing startup ecosystem.
We have become particularly good at building businesses and services on top of technology developed elsewhere.
But there is a fundamental difference between being a consumer of technology and controlling the infrastructure that makes technology possible.
Take a Kenyan bank, for example. Its mobile banking application may have been developed locally. Its payment systems may be homegrown. Its customers may number in the millions. But the underlying computing infrastructure, servers, processors and memory systems ultimately depend on a global hardware supply chain.
The same applies to telecommunications companies, insurance businesses, hospitals, government agencies and virtually every organisation embracing digital transformation.
If memory chips become scarce and more expensive, the consequences will eventually find their way into the cost of doing business in Kenya.
This is not some distant theoretical risk. It is a question of how exposed our digital economy is to decisions made by a handful of semiconductor manufacturers thousands of kilometres away.
And that raises a bigger question: Could Kenya use this moment to move up the technology value chain?
The Microsoft-G42 lesson: Data centres need more than ambition
Consider the proposed $1 billion Microsoft-G42 digital infrastructure investment announced in May 2024.
The initiative was ambitious. It envisaged a geothermal-powered data centre campus in Olkaria, a new Microsoft Azure cloud region for East Africa, local AI research, digital skills development and improved connectivity.
It was potentially a transformative investment for Kenya.
But the project has subsequently encountered significant questions around electricity capacity and implementation. In May 2026, reports indicated that the project had stalled over concerns about Kenya’s ability to supply the enormous amounts of power required. And it is not just a question of availability. As we explored in our own investigation into why Kenyans are still paying dearly for electricity despite the country’s potential for cheaper power, governance failures and structural inefficiencies continue to undermine the affordability of electricity. For a country hoping to attract energy-hungry AI infrastructure and data centres, this is a particularly uncomfortable contradiction: We may have the renewable energy resources to compete, but can we deliver reliable, affordable power at the scale the new digital economy demands?
Citizen Digital’s reporting on the Microsoft-G42 project’s difficulties and Data Center Dynamics’ analysis provide useful context.
There is an important lesson here.
The AI revolution is not simply about software, clever algorithms or buying powerful computers. It requires enormous amounts of electricity, sophisticated cooling systems, reliable fibre connectivity and substantial physical infrastructure.
Geothermal energy gives Kenya a potentially significant competitive advantage. But having abundant renewable energy resources is not the same as having sufficient reliable electricity available at the right location and at the right price.
The World Economic Forum’s July 2026 analysis of Africa’s data centre and electricity requirements estimates that African data centre demand could rise from approximately 0.4GW to between 1.5GW and 2.2GW by 2030.
That is a substantial infrastructure opportunity.
It is also a warning: Countries that fail to prepare their electricity systems could find themselves excluded from an increasingly important part of the global digital economy.
Konza: Could we build more than a Silicon Savannah?
This brings us to Konza Technopolis.
Kenya’s flagship technology city has long been presented as a cornerstone of our ambition to become a regional technology and innovation hub.
There is now an opportunity to give that ambition a much broader industrial dimension.
The development of the Konza National Data Centre and its proposed expansion through a public-private partnership are important steps towards strengthening Kenya’s digital infrastructure.
There is also tangible progress on supporting infrastructure. In June 2026, KETRACO energised the 400kV Isinya-Konza transmission line and the Konza substation, strengthening electricity supply to the area.
These are important building blocks. KETRACO’s announcement provides details of the project.
But perhaps we should be thinking bigger than simply hosting data centres and attracting international technology companies.
Why shouldn’t Kenya explore opportunities in the wider semiconductor and electronics manufacturing ecosystem?
For instance:
- Chip assembly, testing and packaging: These are important stages of semiconductor manufacturing and could offer opportunities to build specialised industrial expertise.
- Electronics manufacturing: Expanding the local assembly of computers, smartphones, telecommunications equipment and other electronic components.
- Hardware research and development: Investing in university-industry partnerships around chip design, embedded systems, robotics and semiconductor engineering.
- Data centre infrastructure: Developing local capabilities in cooling systems, power management, networking equipment and server maintenance.
- Electronic component refurbishment and recycling: Creating a circular electronics industry that recovers valuable materials and extends equipment lifespans.
None of this means Kenya should immediately attempt to build a cutting-edge semiconductor fabrication plant on the scale of Terafab. Such an undertaking would require extraordinary capital, expertise and industrial capacity.
But we do not have to start at the top of the value chain.
Advanced packaging and testing, for example, offer opportunities to develop semiconductor expertise without immediately having to compete with the world’s most sophisticated wafer fabrication facilities.
The bigger question is whether Kenya can develop a deliberate industrial strategy around these opportunities.
Africa’s opportunity – and its vulnerability
There is an even bigger African dimension to this story.
Africa is becoming an increasingly important consumer of cloud computing, digital financial services, artificial intelligence and connected technologies.
Our banks need more computing capacity. Our governments need secure digital infrastructure. Our hospitals need increasingly sophisticated information systems. Our manufacturers will eventually use more robotics and intelligent automation.
All of this requires memory, processors, storage and computing infrastructure.
Yet much of the technology underpinning these services is manufactured outside the continent.
We risk becoming an even bigger market for technology whose most important components we neither manufacture nor control.
The opportunity is to change that equation.
Kenya, Nigeria, South Africa, Egypt, Morocco and other African economies could develop complementary capabilities in data centres, electronics manufacturing, semiconductor design, assembly, testing, technical education and digital infrastructure.
The World Economic Forum’s assessment of Africa’s power and data centre opportunity makes a compelling case for treating digital infrastructure as a strategic economic asset rather than simply another technology investment.
For Kenya, the combination of geothermal resources, an established digital economy, regional connectivity, engineering talent and the ambitions surrounding Konza provides a foundation worth building on.
But the window of opportunity will not remain open indefinitely.
The next technology land grab
Perhaps the most important lesson from Musk’s Terafab venture is that the next generation of technology companies will need to think beyond software.
They will need access to processors, memory, advanced packaging, semiconductor manufacturing capacity, electricity and the infrastructure to support increasingly demanding computing workloads.
The competitive advantage will increasingly belong to companies and countries that secure these resources.
For Kenya, this should prompt a rethink of our technology ambitions.
We have spent considerable time celebrating fintech valuations, startup funding, mobile money innovation and the growth of our digital services industry.
All of that matters. But the next phase of the technology economy may reward countries that invest just as seriously in the physical foundations of the digital world.
We need to ask whether our universities are producing enough semiconductor and hardware engineers. Whether our industrial policies encourage electronics manufacturing. Whether our energy infrastructure can support large-scale computing. Whether Konza can become more than a real estate and data centre project. And whether local businesses can participate in the emerging global hardware supply chain.
We do not need to manufacture the world’s most advanced AI chip tomorrow. But we should be identifying the parts of the semiconductor and electronics ecosystem where Kenya can compete, invest and eventually become indispensable to the African market.
Because while Elon Musk is preparing to build factories to ensure he has enough memory and processors for his robots, Kenya should be thinking about how to build an economy that can supply, support and benefit from the technology revolution those robots represent.
Perhaps the next Silicon Savannah will not be built by writing more code (ChatGPT can do plenty of that these days). It will be built by understanding, manufacturing and controlling the hardware on which the world’s code runs.
And that may be the biggest technology opportunity Kenya has yet to fully appreciate.