Amazon CEO Andy Jassy at AWS re:Invent 2024 (Noah Berger/Getty Images)

super mega ultra

Amazon’s AI plans: custom chips, an Anthropic “ultracluster,” and its own foundation model

While Amazon’s new models appear to be competitive in terms of features and performance, that isn’t the main thing that the company is touting — it’s the cost.

Jon Keegan

12/4/24 2:11PM

This week at Amazon’s AWS re:Invent conference in Las Vegas, the company fleshed out its plans to both serve and compete with the larger AI industry.

AWS is largely AI agnostic. Customers can use pretty much any of the major AI models on the cloud-computing platform, running on servers that use chips from Nvidia, AMD, Qualcomm, and others.

But Amazon has also been building and selling computing powered by its purpose-built AI chips, including its latest Trainium2 chip, which Amazon is now making widely available on AWS’s EC2 service. Amazon says these new Trainium2 instances are built for training and deploying jumbo-sized large language models with better price performance than its current offerings.

Amazon also deepened its partnership with AI startup Anthropic, announcing that it’s building an “ultracluster” of “hundreds of thousands” of Trainium2 servers to train Anthropic’s next-generation LLM. Amazon recently doubled its investment in Anthropic, bringing the total to $8 billion.

Probably the most significant announcement was Amazon’s late entry to the foundational AI-model club. Named “Amazon Nova,” the new LLM comes in four flavors: a text-only Micro and three multimodal models, Lite, Pro, and Premier. Amazon touted benchmark scores for the Nova models, which place it in the same class as OpenAI’s GPT-4o and Meta’s Llama 3. Amazon’s multimodal Nova models can ingest and generate images and videos, like many of the other top models out there today.

While Amazon’s new models appear to be competitive in terms of features and performance, that isn’t the main thing that the company is touting — it’s the models’ low, low cost.

Running Amazon models on Amazon servers, powered by Amazon chips, yields significant cost savings and low latency. Amazon says its Nova models are “at least 75% less expensive” than the best-performing models available on AWS today.

Rani Molla7h

California AG launches probe into xAI and Grok over sexualized deepfakes of women and children

The California attorney general just opened an investigation into xAI, Elon Musk’s AI startup, over chatbot Grok’s apparent role in generating nonconsensual sexual images of women and children. The probe centers on reports that Grok has been used to facilitate the creation of sexually explicit images without consent, many of which have circulated on X.

“The avalanche of reports detailing the non-consensual, sexually explicit material that xAI has produced and posted online in recent weeks is shocking,” Attorney General Rob Bonta wrote in a press release. “As the top law enforcement official of California tasked with protecting our residents, I am deeply concerned with this development in AI and will use all the tools at my disposal to keep California’s residents safe.”

California’s move follows growing scrutiny from US lawmakers and the UK government over AI-generated sexual content and deepfakes.

xAI and Tesla CEO Musk earlier today wrote that he was “not aware of any naked underage images generated by Grok. Literally zero.”

Grok is currently No. 5 on Apple’s free App Store.

Attorney General Bonta Launches Investigation into xAI, Grok Over Undressed, Sexual AI Images of Women and Children

California’s move follows growing scrutiny from US lawmakers and the UK government over AI-generated sexual content and deepfakes.

xAI and Tesla CEO Musk earlier today wrote that he was “not aware of any naked underage images generated by Grok. Literally zero.”

Grok is currently No. 5 on Apple’s free App Store.

Jon Keegan10h

Report: Microsoft on track to spend $500 million per year on Anthropic AI

Last fall, Microsoft and OpenAI’s $13 billion partnership seemed to finally be on solid ground.

OpenAI’s restructuring was completed on time, and the companies hammered out an updated agreement that secured OpenAI’s status as Microsoft’s AI provider of choice, but also allowed for Microsoft to work with other companies.

Now Microsoft is doing exactly that. Microsoft has been increasing its spending on Anthropic’s AI, and is on track to spend $500 million per year on the startup’s services, according to a new report from The Information.

The increasingly cozy relationship between the companies includes the rare move of Microsoft offering incentives to its salespeople that allows Anthropic sales to count toward their quotas, per to the report. Microsoft invested $5 billion in Anthropic as part of a big deal in November that included Nvidia.

Microsoft has also been using Anthropic’s AI to power more and more of its own products, such as Github Copilot and 365 Copilot.

Microsoft’s Spending on Anthropic AI Is on Pace to Hit $500 Million

Microsoft has also been using Anthropic’s AI to power more and more of its own products, such as Github Copilot and 365 Copilot.

Rani Molla11h

Report: Apple staggers Siri AI rollout, with key features pushed to summer

Thanks to Apple’s new partnership with Google, the Gemini-backed version of Siri should begin rolling out this spring, but several key features Apple previewed in 2024 may not come until summer, The Information reports.

The new Siri is soon expected to answer general questions with ChatGPT-like answers — rather than quoting directly from websites or not answering at all. But more personalized, proactive features, like, for example, remembering past conversations and information from them to suggest you leave for a planned trip earlier to beat traffic, may not be unveiled until June at the company’s Worldwide Developers Conference.

The report also clarifies that while Apple’s partnership with Microsoft-backed OpenAI, wherein users could summon ChatGPT for complex questions, isn’t changing, the Google deal might reduce the need for people to do so because Siri will likely be able to answer those questions itself. The Information notes, citing a person familiar with the deal, that the ChatGPT option hadn’t driven much traffic to OpenAI before.

How Apple is Using Gemini to Give ChatGPT-Like Answers

Tesla stops selling self-driving technology as a one-off, pivoting to a subscription model amid slowing vehicle sales

Starting Valentine’s Day, Full Self-Driving will be subscription only.

Rani Molla14h

RIP the metaverse

Meta seems to be winding down its metaverse ambitions. We took a look back at what the company was going for.

Rani Molla, Jon Keegan1/13/26