Chips Galore: AI Competition Heats Up
The fierce rivalry for AI dominance between Advanced Micro Devices (AMD) and NVIDIA is reaching new heights with the unveiling of cutting-edge chips at Computex 2024.
AMD, keen on narrowing the gap with NVIDIA’s stronghold in the AI arena, has been spurred by the escalating digital transformation (DX) of enterprises. IDC projects worldwide spending on DX to soar to approximately $4 trillion by 2027, showcasing a robust CAGR of 16.2% from 2022 to 2027.
According to Gartner, the AI software market is set to witness an explosive CAGR of 19.1% from 2022 to 2027, reaching $297 billion in 2027. Spending on Generative AI (GenAI) software is forecasted to skyrocket from 8% in 2023 to 35% by 2027. Deloitte anticipates enterprise investments in GenAI to surge by 30% in 2024 from $16 billion in 2023.
This voracious spending spree by enterprises signals immense growth prospects for both chip giants.
AMD & NVIDIA: Accelerated Chip Launches
Both AMD and NVIDIA are now zeroing in on swift chip launches to outpace their rivals in the AI landscape.
AMD’s Push Towards Innovation
At Computex 2024, AMD took the wraps off its Instinct accelerator roadmap, showcasing the new Instinct MI325X accelerator slated for release in the fourth quarter of 2024. Featuring 288GB of ultra-fast HBM3E memory, the Instinct MI325X expands AMD’s GenAI performance spectrum.
The upcoming CDNA 4 architecture, expected in 2025, will power the AMD Instinct MI350 Series, promising up to 35X improved AI inference performance compared to the AMD Instinct MI300 Series armed with AMD CDNA 3. The CDNA “Next” architecture will propel the MI400 series accelerators slated for 2026.
AMD introduced the fifth generation of EPYC server processors (Turin) harnessing the Zen 5 core. The company also broadened its lineup with the launch of the Ryzen AI 300 Series, the third iteration of AMD AI-enabled mobile processors, and the Ryzen 9000 Series processors for laptops and desktop PCs.
Noteworthy collaborations were marked at Computex with tech giants like Microsoft, HP, Lenovo, and Asus unveiling new PCs powered by third-gen AMD Ryzen AI 300 Series processors and AMD Ryzen 9000 Series desktop processors.
NVIDIA’s Technological Evolution
Meanwhile, NVIDIA bolstered its array with the introduction of Rubin, the next-generation AI chip architecture set to leverage HBM4. This comes in the wake of the March unveiling of Blackwell, currently in production and slated to reach customers later this year.
The esteemed chip manufacturer plans to roll out new AI chip models annually, an acceleration from the previous biennial timeline. NVIDIA also announced a suite of novel tools and software models at Computex 2024.
NVIDIA’s decision for yearly launches underscores the intense competitive atmosphere in the AI chip market and the imperative to uphold its leading position, currently under siege not just from AMD and Intel but also from innovative strides by Microsoft, Alphabet, and Meta Platforms.
The AI Landscape: A Multifaceted Battle
Microsoft’s Maia 100, tailored for AI operations, stands out with its prowess in running cloud AI workloads such as large language model training and inference. Powered by a 5-nanometer TSMC process and boasting 105 billion transistors, the Maia 100 AI accelerator is a force to be reckoned with.
Alphabet’s Google has thrown its hat in the ring with the Axion central processing unit (CPU), tailored to support its AI endeavors in data centers. These Arm-based Axion processors are devised for robust performance and energy efficiency, offering up to 30% better performance compared to the fastest general-purpose Arm-based instances.
Meta, following Google’s lead, introduced the next phase of the Meta Training and Inference Accelerator (MTIA), a line of custom chips aimed at Meta’s AI workloads. MTIA is set to bolster new generative AI products, recommendation systems, and advanced AI research, bolstered by an 8×8 grid of processing elements delivering enhanced dense compute performance.
In a Nutshell
The meteoric rise of NVIDIA by a staggering 121.4% year-to-date has outshone the growth of AMD, Microsoft, and Alphabet by leaps and bounds. NVIDIA’s stronghold in the AI domain, fueled by high-powered next-gen chips, has made it the top contender for enterprises.
However, AMD’s recent product launches position it as a formidable challenger to NVIDIA, not just in the data center realm but also in the burgeoning AI-enabled consumer PC market in the longer run.
While NVIDIA currently boasts a Zacks Rank #1 (Strong Buy) with an enviable Growth Score of A, offering an enticing investment avenue as per Zacks’ exclusive methodology, its Momentum Score of B further cements its attractiveness to investors.
Conversely, investors eyeing AMD should exercise patience for a more opportune entry point, given its Zacks Rank #3 (Hold) and restrained growth potential in the near term. Moreover, AMD’s premium valuation, sporting a forward 12-month P/E of 40.11X compared to the Electronics – Semiconductors industry’s 32.92X and NVIDIA’s 38.12X, underscores its stretched valuation.