Nvidia (NASDAQ: NVDA) is a stock that could profit from AI. The business pioneered video game GPUs that displayed lifelike pictures. These chips work because parallel processing splits computationally demanding tasks into smaller, more manageable portions. Nvidia became the most successful AI chip manufacturer by adapting this technique to AI.
As demand for AI has increased, many technological giants have been working behind the scenes to develop viable alternatives to Nvidia's industry-leading CPUs. Rivals announced the results of these efforts this week.
1. Intel Intel (NASDAQ: INTC) launched Gaudi 3, its latest AI processor, this week to fight Nvidia's GPU dominance. Inference—running pre-trained AI models—is 50% quicker than Nvidia's H100 flagship data center processor, according to the company. Gaudi delivered all that performance with 40% higher power efficiency than the H100 "at a fraction of the cost," Intel stated.
Remember that Nvidia's Blackwell B200 AI GPU, which will replace its H100, will ship later this year. Gaudi wasn't compared to the B200, but the company said "We do expect it to be highly competitive" against Nvidia's latest chip, calling it "a strong offering."
Instead of outperforming Nvidia's latest AI chips, Intel may offer corporations a cheaper option. The lack of Nvidia's H100 and H200 chips has also slowed AI adoption. Intel's latest product may reduce backlog
2. Meta-platforms To compete, Meta Platforms (NASDAQ: META) introduced the Meta Training and Inference Accelerator (MTIA) v2, a bespoke AI processor for data centers. Meta claimed that its latest MTIA "more than doubles the compute and memory bandwidth" of its predecessor. Early data indicate the new processor outperforms the old one thrice.
Meta's latest chip aims to offer "the right balance of compute, memory bandwidth, and memory capacity". The company isn't processing AI with these chips yet. Meta is employing the CPUs to speed up its rating and recommendation systems, which surface relevant information on social media and target advertising. Meta is one of Nvidia's major clients, so if its AI chip efforts succeed, it might minimize its AI processor dependence.
3. Alphabet Alphabet (NASDAQ: GOOGL) (NASDAQ: GOOG) launched Cloud Tensor Processing Unit (TPU) v5p, its latest AI-centric processor, at Google Cloud Next this week. The business said its Cloud TPU trains AI system-supporting large language models (LLMs) three times faster than version 4. As usual, Google's TPUs won't be sold but will be available on Google Cloud later this year.
Google announced the A3 Mega Supercomputer, which uses Nvidia H100 GPUs, to increase its Nvidia GPU capabilities. Google Cloud customers can also use VMs with Nvidia's Blackwell B200 and Grace Blackwell GB200 GPUs. Google introduced Axion, its first data center CPU, created with Arm Holdings. Axion is 30% faster, 50% more efficient, and 62% faster than Intel and AMD x86 CPUs, according to Google.
stay turned for development