We live in exciting times. Thanks to the internet, starting up, launching a new idea, gathering an early audience, and validating your idea are all possible without spending much time and money. However, the same is not true if you are building a hardware company. Building hardware is still quite expensive; it always has been. But one company that has done a phenomenal job decade over decades has been Intel. Sadly, the last few years have been quite rocky for the once fabled company. In this edition of ROTW, let’s explore what’s happening with the world’s most successful chip designer and manufacturer.

Last week, Intel announced to replace its current CEO Bob Swan; VMWare chief Pat Gelsinger would succeed him. Intel’s problems have been compounding over the last few years, and this is the 3rd CEO change in the last two years.

Intel’s failure over the last decade has been following the very same strategy that has been the reason for Apple’s success during the same time; that is - Integration.

Apple had a phenomenal run in the last 15 years because every product it released (the iPod, iPhone, iPad, or the iWatch), they controlled every aspect of it. Right from manufacturing the custom built hardware to the building differentiated software built specifically to run on that custom hardware. The result was a very profitable business built because of Apple’s strategy to own all parts of the customer experience.

For a very long time, Intel has been following the same strategy. Since the advent of computers in the 70s, Intel started building the first set of microprocessors (8080, 8008, 8086, etc.). It designed the chips required to power the microprocessors, manufactured them, AND built the instruction set (coding language) required to code on these microprocessors.

Intel, by integrating design, manufacturing and software would go on to define and dominate the processor market for decades.

Among these three pillars, the hardest - and the most expensive - part to build, which continues to be the case till today, is manufacturing chips from raw silicon. For a very long time, Intel was the of the very few foundries manufacturing chips from raw silicon. But due to its Apple like vertically integrated approach, it would only manufacture the chip designed by only Intel.

This meant if any company, like Nvidia, for example (who built graphics cards), who wanted to make custom-designed chips would also have to invest in R&D and develop the capability to manufacture those chips. Intel would not manufacture their designs. Building this manufacturing capability would take $100-$150M even in those days, which was very very expensive and didn’t make sense unless you have a massive scale of production.

Seeing this as the opportunity, TSMC (Taiwan Semiconductor Manufacturing Company) setup a pure chip manufacturing company in the late 80s. It would manufacture any custom designed chips built by other companies. Companies like Nvidia, AMD, ARM, etc., benefitted immensely from this because they could now start building hardware without setting up the chip manufacturing unit. They could simply design the chips, and TSMC would manufacture them.

By getting business from many such custom chip design companies, TSMC started producing huge volumes of chips, which led to a lot of innovation and efficiency in their manufacturing processes. As their business grew, they started investing more & more into R&D and perfected the art of building energy efficient chips that are really small in size.

In 2007, when Intel famously decided not to build a chip for the iPhone, Steve Jobs went to Samsung, who could design the chip for the iPhone. Such mobile chipsets would eventually come to TSMC for manufacturing those chips. TSMC had perfected the art of making small energy efficient chips by then, chips that could power mobile phones. As mobile phones took over the world, TSMC (and a couple of other such chip manufacturing foundries) started gaining massive lead up against Intel, which did not have the capability to manufacture small chips for mobile.

Intel still continued to do very well for another ten years because the explosion of mobile phones led to a growth in the cloud technologies, which meant companies like Amazon, Google, Microsoft set up huge data centers to provide cloud services; it meant higher demand for servers. And Intel was still number one in manufacturing chips for servers. Seeing the mobile revolution, they did invest in building small chips, but they were far behind the likes of TSMC in their ability to make great quality chips for mobile, that they could never get a customer.

And then came the final blow for Intel. In the last few years, as cloud computing companies have consolidated into a market of three (Amazon, Google & Microsoft), these companies have started investing in designing their own chips, thanks again to the likes of TSMC, who can manufacture any designs.

The thing about processor chips for mobile is that you need the smallest possible chip that requires the least amount of power because billions of such chips put together on a small circuit board make a processor. To draw a comparison, the Intel chip used in Apple laptops is 45nm (nanometres, 1nm = 0.000000001 meters). The TSMC chip in Qualcomm Snapdragon 800 (one of the successful mobile phone processors) was 7nm. And it was manufactured a couple of years ago. Today, TSMC is trying to build a 3nm chip, while the smallest chip that Intel can produce is 10nm; their 7nm production is struggling massively (should take then three more years according to their own estimates; by that time, TSMC would have started produced 2nm chips).

This leaves Intel in a very bad place. It can’t build small chips, and the business in which they were the market leaders - cloud servers & PCs - is either getting shrunk or getting disrupted with their clients building their own chips.

Here is the thing. When you are at the top, there is only one way you can go. Even to stay where you are, takes tremendous effort and foresight. Intel could have chosen to part with some of its profits and invest it into R&D and develop these smaller chips sooner. Instead, it chose to double down on its most profitable product line - the bigger and more powerful server chips. That strategy continued to pay dividends right up to the point till it didn’t.

Bog Iger, CEO of Disney, once said, “The riskiest thing we can do is just maintain the status quo.”

Intel, to its own peril, did precisely that.

Best,
Kaddy

The Invisible Rules That Hold Us Back

On Outworking Your Competition