If I suggested we were investing in semiconductor companies, most people would immediately picture Nvidia. Quite understandably so. Over the past two years, the company has become almost synonymous with artificial intelligence, to the point where ‘AI exposure’ and ‘Nvidia’ are often treated by markets as near interchangeable concepts.
The reality, however, is that the semiconductor industry is vastly broader, more nuanced and, frankly, more interesting than that.
Whilst Nvidia has come to dominate headlines, the semiconductor ecosystem spans everything from hyperscale AI infrastructure and networking through to industrial automation, automotive systems, medical equipment and factory machinery.
Companies may all technically ‘make chips’, but the underlying economics, customer bases and long-term investment characteristics can look entirely different depending on which part of the market they operate in.
That’s part of the reason we’re comfortable holding exposure to businesses as different as Nvidia, Broadcom and Texas Instruments within portfolios at the same time.
Nvidia has, quite understandably, become the defining company of the current AI investment cycle. Its graphics processing units sit at the centre of many of today’s most advanced AI models and the company has become one of the clearest beneficiaries of the enormous capital expenditure currently flowing into data centres and computational infrastructure.
Broadcom occupies a slightly different but equally important role within that ecosystem, benefiting from both networking infrastructure and the growing demand for custom silicon as the largest technology companies increasingly seek to optimise their own AI capabilities.
These are businesses operating at the frontier of computational intensity and digital infrastructure. Their growth profiles have been exceptional, but so too has the level of market excitement surrounding them.
Texas Instruments, by contrast, represents a rather different side of the semiconductor world. The company specialises primarily in analogue and embedded semiconductors; the less glamorous but deeply essential components that help the physical economy continue functioning.
Its chips are used across industrial equipment, factory automation, automotive systems, power management, communications infrastructure and medical devices, often performing highly specific tasks over very long product lifecycles.
In many ways, Texas Instruments reflects the quieter reality of semiconductors. Whilst the market focus naturally gravitates toward artificial intelligence and cutting-edge computing power, much of the global economy still depends on relatively simple chips performing repetitive but mission-critical functions reliably and efficiently.
A factory production line, for example, has little interest in whether a semiconductor can generate poetry or create photorealistic images; it simply needs systems to continue operating safely and consistently.
One business is helping train large language models capable of answering existential philosophical questions at extraordinary speed. The other is making sure your car braking system, factory conveyor belt and air conditioning unit continue behaving themselves. Financially speaking, though, both can be extremely attractive businesses for very different reasons.
Although these companies all sit broadly within the semiconductor universe, they’re exposed to very different drivers. Nvidia and Broadcom are more closely linked to AI investment cycles, hyperscaler spending and the rapid expansion of computational demand.
Texas Instruments is more closely tied to industrial production, automotive content growth and the gradual digitisation of the physical economy. One benefits from explosive technological acceleration, the other from durability, breadth and long-cycle industrial demand.
We don’t necessarily view those exposures as competing with one another. If anything, they can be highly complementary within a long-term portfolio.
One of the persistent challenges during periods of technological excitement is that markets often compress nuance into a single narrative. Entirely different businesses can begin trading as shorthand for the same theme, despite having very different competitive positions, customer bases and risk profiles.
Investors saw similar behaviour during earlier phases of the internet buildout, cloud computing and electric vehicles. Artificial intelligence has merely accelerated the tendency.
In reality, technological revolutions rarely produce a single winner. They tend to create multiple layers of beneficiaries operating across different parts of the value chain, each with distinct characteristics.
For investors, that’s where portfolio construction becomes interesting. Long-term investing isn’t simply about identifying important themes, but about understanding where sustainable economics sit within those themes and how diversified sources of return can coexist alongside one another.
Eleanor Ingilby is head of high net worth at Atomos. The views expressed above should not be taken as investment advice.