In a $40 billion deal, NVIDIA has agreed to buy Arm from Softbank and the company promises to preserve Arm's independent business model while trying to expand its offerings with AI and GPU IPs. When the speculation of NVIDIA's acquisition of Arm was swirling around, we did a piece analysing the implications of the deal.  (NVIDIA-Arm: A Wild Idea or Could it Work?)

With this acquisition, NVIDIA gets access to one of the most successful ecosystems ever created in the technology industry. Over the last 25 years, Arm has built a super successful IP licensing business and served as an R&D outsourcing shop for the semiconductor industry. NVIDIA is known for GPU-based AI computing prowess while Arm has made waves in the low-power computing devices, especially in smartphones. While there are some common areas between the two companies, by and large, both companies have highly complementary assets.

While this deal provides an excellent exit deal for Softbank, NVIDIA will have to work hard in the next 24 months to convince the Arm ecosystem partners, governments and shareholders as to why this is a right move. From a financial point of view, Arm makes a good business case given its highly profitable recurring revenue streams from IP licensing business. But its revenue growth has been sluggish in recent years.

Under Softbank's ownership, Arm expanded into new markets including IoT, security, automotive, networking and data centre but the company faltered in areas such as AI. Even its core mobile CPU and GPU businesses came under pressure, especially with reduced demand for smartphones in recent years.

Will NVIDIA do any better than Softbank? Between 2015 and 2019 NVIDIA more than doubled its revenue while Arm grew its revenue just 5 percent CAGR during the same period. NVIDIA's high growth can be attributed to its leading position in the data centre AI platform market, which saw massive growth.

With enough concessions, NVIDIA may get regulatory approvals for this acquisition but such concessions could defeat the very purpose of the acquisition. NVIDIA thinks that its Mellanox regulatory experience could be handy but we feel that Arm is an entirely different asset and may face more scrutiny.

Beyond regulatory hurdles, some of the pertinent questions include:

• Will Arm's leading licensees approve NVIDIA's ownership and influence on the critical IP of their devices and chips? Our view is that NVIDIA's ownership of Arm will not be viewed favourably by some Arm licenses including Qualcomm. Qualcomm had to pay over $270 million in fine as a result of Icera's (NVIDIA's modem division) complaint to the European Commission.
• Will co-opetition work here? We note that co-opetition (companies that compete in one area also partner in another) has been a common practice in the smartphone industry. For example, Apple, Huawei and Samsung buy modem chips from Qualcomm while competing with their apps processor products.
• What is the immediate alternative to Arm licensees? The good news for NVIDIA is Arm's licensees do not have a close choice to change the course and likely stick with NVIDIA-Arm in the next 3-5 years. The competing architecture RISC-V, while used in MCUs, flash memory controllers and RF processing, is not yet ready for smartphone SoCs and could take another 4-8 years to be smartphone ready. Software support is a significant barrier for RISC-V taking-off in smartphone SoCs. We note that some RISC-V companies such as Si-Five have a head start and might become acquisition targets.
• Can NVIDIA return to smartphone SoCs? Our view is that this move is unlikely, at least in the next three years. Between 2010 and 2014, NVIDIA briefly participated in the smartphone SoC market with its Tegra chips and ended up winding down its modem division Icera in 2015. We think modem technology is crucial to stay competitive in the smartphone SoC market and both Arm and NVIDIA currently lack 5G modem technology. NVIDIA has already tried to license its GPU cores and Deep Learning Accelerator IPs to smartphone vendors without much success. NVIDIA may explore the PC CPU market with Arm designs. This market is set to be more competitive with Apple's entry with its Apple Silicon.
• Can NVIDIA differentiate itself enough while licensing its critical AI and GPU IP to competitors?
• The PR suggests that Softbank exclusively negotiated with NVIDIA and there was no other bidder. Companies with vested interests in Arm ecosystem may likely form a consortium to thwart NVIDIA's bid or those companies just wait for NVIDIA to go through a two-year regulatory hurdle and fail.

NVIDIA's grand vision to dominate from data centre to edge with Arm acquisition could create ripples in the industry. Until recently, NVIDIA sat on the side-lines when it comes to large acquisitions and the company didn't participate in the peak semiconductor consolidation period. Arm has the right to be purchased but the ecosystem appears to be reluctant to a competitor-owned Arm.