I was reminded of this quote today with news that Amazon will invest $4 billion in Anthropic, the San Francisco-based startup behind the Claude 2 AIGC chatbot and a major competitor to OpenAI and its ChatGPT.
In "Star Wars: Episode II: Attack of the Clones," a young Jedi Master, Yoda, utters a meme-worthy line in his signature: The Clone Wars have begun.
I was reminded of this quote today with news that Amazon will invest $4 billion in Anthropic, the San Francisco-based startup behind the Claude 2 AIGC chatbot and a major competitor to OpenAI and its ChatGPT.
However, as far as the technology industry is concerned, the sentence is more like "the AI war has begun." Cloud computing leaders Amazon, Microsoft and Google are now supporting different AI basic models and developing their own models.
Of course, if you've been paying attention to tech news coverage of AI over the past few months or years, you're certainly aware that Amazon is far from the first household name to venture into the AIGC space. AIGC refers to AI that uses ML algorithms to generate new data after training on a large amount of previous data.
In fact, Microsoft invested in an AI startup called OpenAI in 2019, and again in 2021 and earlier this year, committing billions of dollars. This is Microsoft's true beginning.
Not to be outdone, Google announced a $300 million investment in Anthropic in February, a figure that now seems minuscule by comparison. The search giant is also backing video AIGC startup Runway ML.
Still, today's news of Amazon's big investment shows that competition among the three major cloud providers (by market share) has escalated, and they will work with them to benefit and offer their consumers the latest and greatest. AIGC technology.
In a swift move, Amazon has replaced Google as Anthropic's primary technology backer, gaining a potentially extremely valuable ally as demand and interest in AIGC continues to rise across various industries.
Why is this moment qualitatively different from the investments and announcements made in the AIGC space to date? I can offer several reasons:
In the wake of the cloud war, big tech companies see AI as the next and most important frontier (the two are inseparable).
When most people think of cloud giants, their popular consumer-facing services probably come to mind: Microsoft makes computers and software, Amazon sells products on the Internet, and Google lets you search and send email.
Before the advent of AI, however, the most recent—and arguably most important—competition between Microsoft, Google, and Amazon was the battle for cloud computing.
With so many of us as individuals and our companies generating and relying on data, the cloud has become more important than ever as a place to not only store this data, but also run applications on it, including, of course, generative AI app.
According to market research firm Gartner, the global cloud services market is expected to grow from US$491 billion last year to US$597.3 billion this year, a year-on-year growth of 21.7%, which is largely driven by the AIGC boom.
When Gartner Vice President Analyst Sid Nagle released a cloud service forecast report in April 2023, he said: "For example, AIGC is supported by large language models, which require powerful and highly scalable computing power to Process data in real time.” The cloud provides the perfect solution and platform. It's no coincidence that the key players in the AIGC race are cloud superscalers.
Current shortcomings may explain rush to embrace AI
Amazon's cloud business, AWS, contributed nearly 70% of the company's profits last quarter. According to Gartner, as the world's leading cloud service provider, Amazon has nearly 40% market share and is the leader in cloud computing.
At Microsoft, Satya Nadella was promoted to CEO in 2014 after Steve Ballmer lost out to Apple and Google. Nadella was chosen for his success in running Microsoft's cloud and enterprise division, which was responsible for the launch of Microsoft Azure.
Today, Microsoft Azure is the second-largest cloud provider behind AWS and closing in on it—a respectable position, but you can bet Microsoft would rather be first. It would be even better if AIGC helps the company do this by increasing demand for AIGC companies like OpenAI.
Amazon Alexa isn't enough
Meanwhile, Amazon's move to support humans appears to be a smart one, as the company just last week announced its new Alexa assistant powered by large language models.
The timing of the announcement is odd considering Amazon is an early leader in conversational AI assistants, having launched Alexa as a voice for its first Amazon Echo device back in 2014. Shouldn’t Amazon already have all the necessary internal data and expertise to build a competitive AIGC model?
While Amazon's Alexa is in some ways the most popular voice assistant, the smart speaker market it helped launch has begun to stagnate, and competition in the space has only intensified significantly in the past decade as Apple paired its HomePod with Bundled with the Siri voice assistant, Google offers the Google Assistant on various Nest and Google Home devices. Amazon's new CEO Andy Jassy, who took office in 2021, has laid off employees in Amazon's device division. Reuters recently reported that the team in the division is said to have low morale and a weak development pipeline.
With its current device strategy facing headwinds and the release of the first large Alexa language model nearly a year after OpenAI debuted ChatGPT — which, by all accounts, is a more capable and powerful AI assistant — Amazon appears to be Realize that there's still a lot of work to do to catch up in AI, and see Anthropic as a good shortcut, even if it might undercut AWS's "Swiss" positioning in running AI models and storing training data.
Of course, weakening its platform-neutral role has never stopped Amazon before: The company's Amazon Basics line of products competes with other products from leading third-party vendors in its e-commerce marketplace, and its Amazon Prime Studios unit produces movies and TV shows, Compete for viewers' attention with leading studios whose films are also available to rent, buy, or stream for free through Amazon Prime Video (and free video).
Google is catching up
Finally, we come to Google. The company essentially launched the second generation of the AI revolution of the past five years after some researchers at its AI subsidiary Google Brain published the seminal 2017 paper "Attention is All You Need," outlining An open-source approach to building such transformer models that power OpenAI’s ChatGPT, Anthropic, and nearly all of the leading consumer and enterprise AIGC models popular today.
But Google's vast, notoriously slow and politicized internal bureaucracy failed to retain those researchers, who have since created a new generation of AI startups that the company is now competing directly with, including Cohere and Sakana AI.
As Michael Nuñez noted in commenting on the latest update to Google Bard, the search giant's AIGC assistant based on its large language model for the Palm 2, Google's AIGC efforts have so far failed to impress, competing in terms of functionality and utility. OpenAI pales in comparison. No wonder Google is reportedly moving quickly to release a true competitor to OpenAI’s underlying AIGC model GPT-4, which it calls “Gemini.”
Meanwhile, Google is trying to climb from a distant third place behind AWS and Microsoft in the rankings of cloud service providers. On both fronts, Google needs a big win: a powerful new generation of AI models could be the proverbial kill of two birds with one stone.
Google may plan to leverage some of Anthropic's technology to help achieve this goal, but with Amazon sweeping the market as its new, higher-return backer, it looks like Google will have to deploy a compelling new AI model alone, and Making its cloud the preferred back-end service for a new generation of AI models and applications.
where does this leave us
Although the one-year anniversary of ChatGPT’s launch in November is approaching, the new generation of AI wars has only really begun. If the game were a baseball game, we'd be in the first few innings. With the pace of competition and spending increasing so rapidly, the game will likely continue to be played out for many more rounds.
Now, unlike games, in this new generation of AI, there won't necessarily be a "winner takes all" situation. Indeed, projections of the technology's potential applications and its coverage clearly support a scenario in which there are multiple technology providers, some large and some small, some specialized and some general.
Of course, we cannot ignore the importance of open source AI. Currently, Meta Platform leads the way among existing technology giants by open source and licensing commercial use of its Llama 2 language model.
Llama 2 and other open source models, such as Falcon180B and DeciLM6B, provide a "level playing field" approach, allowing any enterprise to access the powerful AIGC tools to build applications, or modify them for their specific use.
However, even in this case, those deploying and fine-tuning open source AI models still need cloud servers to store data and run inference, which is why the three major cloud players - Amazon, Microsoft and Google - also have incentives to provide them themselves or through collaborations The preferred AI products provided by partners.
If you are an enterprise administrator supporting Microsoft Azure or AWS, why not use their recommended AI models and instead try to build or run your own? This seems to be what is driving the next generation of collaborations and strategies in the AI/cloud space .
Whatever happens, what is certain is that the war will be a fierce and costly one.