Arm can make Nvidia the most important chip maker

In the previous post, I discussed that Nvidia’s deal for Arm is very smart because it was structured as a majority stock acquisition. In this post, I will discuss the strategy driving the acquisition and how Arm will help make Nvidia the most important chip maker in the world.

Last July, Nvidia surpassed Intel to become the largest U.S. chip maker by market value. However, by most measures, Nvidia remains a niche player; compared to Intel’s $72 billion in revenue last year, Nvidia’s sales were a little over $11 billion. So how can Nvidia grow to its market value? Nvidia thinks the answer is Arm.

This answer is not obvious to everyone, though, because the two companies’ business models couldn’t have been more different. While Nvidia sells vertically integrated solutions and doesn’t share its IP, Arm sells modular designs that are available to everyone in the ecosystem. 

• Nvidia started as a niche player and made its name making and selling best in class graphic processor units (GPU). Nvidia integrates and differentiates its premium hardware (GPU) with a software platform called CUDA that abstracts the complexities of graphics programming. CUDA is available for free but here’s the catch: it only works with Nvidia GPUs. CUDA is dominant in machine learning and developers have no choice but to pay a big premium for Nvidia’s GPUs.
• Arm, on the other hand, does not make any chips of its own. It licenses chip designs to companies to manufacture them and collects a licensing fee. Today, nearly every smartphone in the world runs on an ARM-derived chip. This happened because Arm’s designs are more power-efficient than Intel's. Unlike the CUDA platform, Arm’s license is not free but it’s neutral - for a fee anyone in the ecosystem gets equal access to Arm’s designs. 

Why does Nvidia need to own Arm, instead of simply licensing it? 

Despite all the devices that Arm’s chips are on (e.g. smartphones) and the potential for many more (wearables, IoT etc.), it can't be that Nvidia acquired Arm for the pennies in licensing fees. Nvidia's entire business is predicated on selling high margin chips differentiated by highly integrated software. So why trade high margins at the top of the market for low margins in licensing fees and the headache of serving everyone?

Nvidia has its sights firmly set on Intel and its data center business. While Intel has leveraged its integration of design and manufacturing to dominate, Nvidia is going to leverage its integration of chip design (Arm) and software (Nvidia’s CUDA). Nvidia believes that it’s the lack of a software platform that is holding Arm in the data center - and CUDA is the answer. When Nvidia’s CEO says Nvidia itself is going to adopt Arm’s business model of licensing its IP to competitive chip-makers, he means licensing chip designs from Arm and CUDA from Nvidia. 

This is exciting and assuring for ARM licensees (e.g. Amazon, which has invested in Arm chips for the data center), many of whom are probably worried that Arm’s neutrality under Nvidia is at risk, if not gone. After all, Nvidia is known to be very difficult to partner with. It’s important to highlight though that Nvidia isn’t doing this out of charity. Nvidia is intent on capturing the upside that CUDA will generate, which means ever increasing licensing fees; it will still be cheaper than Intel’s costs.

If this vision is successful, acquiring Arm will have huge upsides for Nvidia. Arm will have a great opportunity to knock Intel out of another business. On the surface, this deal looks like Nvidia is adopting Arm’s business model. In reality, Nvidia is sticking to its business model of vertical integration. Nvidia will sell high-margin designs differentiated by highly-integrated software in data centers. I like where this is going.


Disclaimer: This post is merely my own assessment and is not an investment recommendation. For professional advice, seek input from a licensed investment advisor.