Amazon wants chips in everything. It is creating its own AI-focused silicon to dethrone Nvidia

When I think of Nvidia, I think "games," but that's not the whole picture. If we're just talking about cold, hard cash, AI is becoming a massive source of revenue, bringing in $30 Billion during Q2, and far out-shining the share of revenue that games brought in, totaling $2.88 Billion during the same period. Nvidia, a major player in this field, is no doubt a big-hitter. But there's about a rumble to come in the AI jungle.

Amazon has never been shy about its AI commitment (and we're talking about more than just their robot that spits out deals). The company is not only investing in nuclear power to power their electric dream, but also aiming for the top when it comes AI chips (via Financial Times).

Amazon wants to reduce its reliance on Nvidia and is pooling all their chips in order to produce their own. Amazon hopes that its own-made chips will increase the efficiency of Amazon's data centres and reduce their operating costs. This is good news for Amazon, but it could also benefit customers of Amazon Web Services.

Where is Amazon spending money? Annapurna Labs is the answer. Amazon bought this start-up in 2015 for $350m. Together, they've been working on the 'Trainium 2 AI chip' for at least the past year. It looks like I won't have to wait long to get my hands on it.

Anthropic is already putting these chips to the test. Anthropic, a competitor to OpenAI and another startup that has received generous financial backing (to the tune $4 billion) from Amazon among other investors.

Amazon is also working on a new line of AI chips called 'Inferentia', which it claims is already 40% more cost-effective in terms of generating AI responses. It's a much better name than Trainium, which means 'inference in Latin'. However, I hope they don’t follow up with chip lines named Suspiria or Tenebrae (although a horror director Dario's return to AI could be interesting).

Amazon's latest AI bid, despite its giallo-daydreams, is not the only one. Microsoft and Meta are also looking to unseat Nvidia, by creating their own chips that better meet AI demands in their respective businesses. This is already a big move in the tech ocean, but there are more major players likely to follow.

AI is a big money maker for Nvidia, so it's not surprising that other companies are trying to get a piece of the pie. These companies probably hope for continued growth, but that may be wishful thinking. OpenAI's founder believes that large language model learning has reached a plateau. This could mean the bubble will burst sooner than we expect.