Back to Innovator’s Dilemma. Three decades ago, ARM started off as an innocent snowflake. Today, ARM is thundering like an avalanche that could potentially break into Intel’s moats by the mountainside. It is text book disruption, as Prof. Christensen described it.

The question for this blog is, what could disrupt ARM’s position over the next decade or more? While its designs power millions of devices, ARM is still a sub-scale ~$2B company. Its masters (Softbank or Nvidia, as the future may be), bought it for its potential financial returns – they will seek to realise it. They will not be just contented with its strategic benefit – like Oracle stayed with Sun Java.

In the 80s and 90s, Intel disrupted IBM and other architectures. ARM was but a trifling snowflake. By the 2000, x86 looked like it was the master of all it surveyed. But by 2020, we know that is not the case. The snowflake rolled into an avalanche and broke into Intel’s moats.

Now, how could ARM get disrupted in the eternal cycle of life? Say by 2035? This is like predicting which snowflake will roll itself into a dam ripping avalanche. ARM’s current position has been manoeuvred over three decades. It has some way to go before it declares success. Its displacement, if at all it happens, might take a long time.

With a light business model and no manufacturing, ARM is more anti-fragile than Intel is. But with a small revenue book and profitability, it is not showing all the characteristics of a ‘moat’.

What are the forces in the distant future that will disrupt ARM? In 5 years? May be in 20 years? Who knows? Why would it happen? Let us speculate gleefully.

This is an exercise of distilled, unabashed, unbridled, uninformed, joyful speculation.

Here are a few scenarios I could imagine:

• Nvidia deal doesn’t go through. Softbank still shops around ARM. Being kicked between different owners at shrinking valuations, ARM runs out of cash. This very unlikely as the licensees have the interest in keeping the company up and running.
• Nvidia takes over and changes licensing terms, forcing the aggressor Qualcomm or Apple (If Nvidia takes over and changes the reality, these two have the maximum to lose) to fork RISC-V. In this scenario, ARM dies in the court between the litigious Apple/Qualcomm and combative Jensen Huang (Nvidia CEO).
• IBM open sourced the Power’s instruction set in 2020. They even open sourced the design of the chip that powered the Big Gene. Will companies like HP, Dell take up this instruction set to run their PCs? Will IBM evangelise with the devices eco-system to drive adoption of Power standards?
• What about RISC-V open source instruction set? It serves the same function as ARM instruction set. It is supposed to be power efficient and offer similar levels of performance. And it is completely open source. Will it be adopted by a ‘outsider snowflake’ – Elon Musk for example, to drive a long term project for open source processor architecture – for space rockets for example?
• Will the technical eco-system get splintered due to geo-political risk? In a bad version of the future, there might be tight embargo from the West to China. May be China and its allies go into a different architecture while the rest of the world goes on with x86 and ARM. Once the Iron Curtain breaks again, we find China’s architecture to be far better and adopt it. [Just like US and USSR had different space ‘architectures’. US used the Russian space program until recently to travel to the ISS]. What if China shuns the Western technology standards, and Chinese standards turn out to be more pervasive?
• May be in another version of the future, Microsoft or Google or Amazon (or the like) might want to dominate Robotic OS [Replete with postronic codes embedded with the holy laws] – they develop and open source an advanced processor architecture for Robot manufacturers.

Intel has maximum to lose in this game. May be Intel catches up on 3nm or better. Intel decides the value in microprocessors is in the manufacture – not in the design. Intel then puts in a team to develop on RISC-V (or some alternative) architecture commercialises it. It sustains the team from licences, customisation and manufacturing revenues from its FABs.

In this version of the future, Intel’s considers TSMC’s profits! as its opportunity. This is a very plausible scenario, given Intel’s engineering capabilities.

• May be by 2035 quantum computers take over the data centre and shifts a large % of compute to the cloud – all powered by green tidal power. Just like Echo/Alexa architecture shifted computing to the cloud, may be it might be better in that state to shift PC/Smartphone computing also to the cloud. Yeah, at that time we have 12G or something. Then, this entire discussion might become irrelevant as the devices are just dumb clients to the master server.

It is most likely that there will be another, currently unimaginable, hidden snowflake that causes the next disruption. And then the great Prof. Christensen will smile from his grave.

History, like a broken tape, will unrelentingly repeat itself.

Needless to say, which ever the scenario, computing as we know will get better. We will get more out of our devices. Life will be interesting.

Food will not be teleported to your kitchen though. It might be delivered by ARM powered drones though.

But you can call Mom home through the Haloport.

The world will be a better place.

! With ARM’s revenue at $2B [Loss of $400M], and TSMC at $45B[Gross Profits of $24B], it is clear that chip design is NOT capturing the value. Manufacturing is.

In one version of the future, Intel can consider TSMC’s profits as its opportunity.

Next: Written on Rosetta Stone #15: Summary