Key Points

• 1 As Microsoft and Google take the lead in the field of generative artificial intelligence, Amazon is also catching up. It is reported that the company has secretly designed two chips in Austin, Texas, for training and accelerating generative artificial intelligence.
• 2 The two custom chips developed by Amazon, Inentia and Trainium, provide an alternative option for AWS customers, replacing the increasingly difficult to procure Nvidia GPUs for training large language models.
• 3 AWS's dominant position in cloud computing is a major advantage for Amazon. AWS is the world's largest cloud computing provider, occupying 40% of the market share in 2022.
• 4 Analysts believe that in the long run, Amazon's custom chips may give it an advantage in the field of generative artificial intelligence.

In an unremarkable office building in Austin, Texas, several Amazon employees are designing two microchips in two small rooms for training and accelerating generative artificial intelligence. The two custom chips, codenamed Inentia and Trainium, provide an alternative option for AWS customers, replacing Nvidia graphics processors for training large language models. Currently, it is becoming increasingly difficult to procure Nvidia graphics processors, and their prices are rising.

Adam Selipsky, CEO of AWS, stated in an interview in June: "The world is looking for more chips to support generative artificial intelligence, whether it's graphics processors or chips designed by Amazon itself. I think, compared to any other company in the world, we are more likely to provide this capability that everyone wants for our customers."

However, other companies have taken faster action, invested more funds, and gained momentum from the artificial intelligence boom. When OpenAI launched ChatGPT in November last year, Microsoft gained widespread attention for hosting this popular artificial intelligence chatbot. It is reported that Microsoft invested $13 billion in OpenAI. In February of this year, Microsoft quickly added generative artificial intelligence models to its products and integrated them into Bing.

In the same month, Google launched its own large language model, Bard, and then invested $300 million in OpenAI's competitor, Anthropic.

It wasn't until April of this year that Amazon announced its own large language model, Titan, and also launched a service called Bedrock to help developers enhance software capabilities using generative artificial intelligence.

Chirag Dekate, Vice President and Analyst at Gartner, said: "Amazon is not used to chasing the market, but is used to creating the market. I think, for a long time, they have found themselves at a disadvantage for the first time and are now working hard to catch up."

Meta recently released its own large language model, Llama 2, and this open-source competitor of ChatGPT can now be tested on the Microsoft Azure public cloud.

Chips Represent "Real Differentiation"

Dekate said that in the long run, Amazon's custom chips may give it an advantage in generative artificial intelligence. He explained: "I think the real difference lies in the technological capabilities they possess, because Microsoft does not have Trainium or Inentia."

Image: AWS has been producing custom chips Nitro since 2013, which is currently the largest capacity AWS chip.

As early as 2013, AWS quietly began producing custom chips with a dedicated hardware called Nitro. Amazon revealed that Nitro is currently the largest capacity AWS chip, with at least one on each AWS server, totaling over 20 million in use.

In 2015, Amazon acquired Israeli chip startup company Annapurna Labs. Then in 2018, Amazon launched the server chip Graviton based on the Arm architecture from the UK chip design company, which competes with giants like AMD and Nvidia's x86 CPUs.

Stacy Rasgon, Senior Analyst at Bernstein Research, said: "The proportion of Arm chips in total server sales could be as high as 10%, with a large part coming from Amazon. So, in terms of CPUs, they are doing quite well."

Also in 2018, Amazon launched chips focused on artificial intelligence. Two years ago, Google released the first Tensor Processor Unit (TPU). Microsoft has not yet announced its collaboration with AMD to develop the Athena artificial intelligence chip.

Amazon has a chip lab in Austin, Texas, where Trainium and Inentia are developed and tested. Matt Wood, Vice President of Products at the company, explained the uses of these two chips.

He said: "Machine learning is divided into these two different stages. So, you need to train machine learning models, and then infer these trained models. Compared to other ways of training machine learning models on AWS, Trainium improves cost-effectiveness by about 50%."

Following the launch of the second generation Inentia in 2019, Trainium was first launched in 2021. Wood said that Inentia allows customers to "provide low-cost, high-throughput, low-latency machine learning inference, which is all the predictions you get when you input prompts into generative artificial intelligence models, all of which are processed, and then you get a response."

However, for now, Nvidia's graphics processors still dominate in training models. In July of this year, AWS launched new AI acceleration hardware based on Nvidia H100.

Rasgon said: "Over the past 15 years, Nvidia has built a huge software ecosystem around its chips, which other companies do not have. Currently, the biggest winner in the field of artificial intelligence is Nvidia."

Image: Amazon's custom chips, from left to right: Inferentia, Trainium, and Graviton.

Amazon Has Cloud Computing Advantage

However, AWS's dominant position in cloud computing is a major advantage for Amazon.

Chirag Dekate, Vice President and Analyst at Gartner, said: "Amazon doesn't need to pay extra attention, the company already has a very strong cloud installation base. What they need to do is figure out how to use generative artificial intelligence to help existing customers expand into value creation."

When choosing generative artificial intelligence between Amazon, Google, and Microsoft, millions of AWS customers may be attracted to Amazon because they are already familiar with Amazon, running other applications there and storing data.

Mai-Lan Tomsen Bukovec, Vice President of AWS Technology, explained: "It's a matter of speed. These companies' ability to develop these generative artificial intelligence applications at what speed depends on starting with data from AWS and using our computing and machine learning tools to drive it."

Data provided by Gartner shows that AWS is the world's largest cloud computing provider, occupying 40% of the market share in 2022. Although Amazon's operating profit has declined year-on-year for three consecutive quarters, AWS still accounts for 70% of Amazon's $7.7 billion operating profit in the second quarter. Historically, AWS's operating profit margin is much higher than that of Google Cloud.

In addition, AWS has an increasingly comprehensive developer toolset focused on generative artificial intelligence. Swami Sivasubramanian, Vice President of Databases, Analytics, and Machine Learning at AWS, said: "Let's turn the clock back, even before ChatGPT. This didn't happen after that, we suddenly rushed out a plan, because you can't design a new chip in such a short time, let alone build a basic service in 2 to 3 months."

Bedrock allows AWS customers to access large language models developed by Anthropic, Stability AI, AI21 Labs, and Amazon Titan. Sivasubramanian said: "We don't believe that one model will rule the world, we want our customers to have the most advanced models from multiple vendors, because they will choose the right tools for the right job."

Image: Amazon employees are developing custom artificial intelligence chips at the AWS chip lab in Austin, Texas.

One of Amazon's latest artificial intelligence products is AWS HealthScribe, which was launched in July to help doctors draft patient visit summaries using generative artificial intelligence. Amazon also has the SageMaker machine learning center, which provides algorithms, models, and other services.

Another important tool is CodeWhisperer, which Amazon says has increased the average speed of developers completing tasks by 57%. Last year, Microsoft also reported that its coding tool, GitHub Copilot, improved work efficiency.

In June of this year, AWS announced a $100 million investment to establish a generative artificial intelligence innovation center. AWS CEO Selipsky said: "Many of our customers want generative artificial intelligence technology, but they may not necessarily know what it means for them in their own business context. So, we will introduce solution architects, engineers, strategists, and data scientists to work with them one-on-one."

CEO Jassy Personally Leads the Construction of Large Language Models

Although AWS has mainly focused on developing tools rather than creating competitors to ChatGPT, a recently leaked internal email shows that Amazon CEO Andy Jassy is directly overseeing a new central team, which is also building scalable large language models.

In the second quarter earnings call, Jassy stated that "a significant portion" of AWS business is now driven by artificial intelligence and its support for over 20 machine learning services, with clients including Philips, 3M, Old Mutual, and HSBC.

The explosive growth of artificial intelligence has raised a series of security concerns, with companies worried that employees will input proprietary information into the training data used by public large language models.

AWS CEO Selipsky said: "I can't tell you how many Fortune 500 companies I've talked to that have disabled ChatGPT. So, anything you do through our generative artificial intelligence approach and Bedrock service, any model you use, will be in your own independent virtual private cloud environment. It will be encrypted, and it will have the same AWS access controls."

Currently, Amazon is only accelerating the advancement of generative artificial intelligence, claiming that "over 100,000" customers are using machine learning on AWS. While this is only a small portion of AWS's millions of customers, analysts suggest that this situation may change.

Chirag Dekate, Vice President and Analyst at market research firm Gartner, said: "We haven't seen enterprises saying, 'Oh, wait, Microsoft is ahead in generative artificial intelligence, let's go out and change our infrastructure strategy, let's move everything to Microsoft.' If you're already an Amazon customer, you're likely to explore Amazon's ecosystem more broadly." (Text/Golden Deer)

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。