Can Amazon and Meta catch up to Microsoft and Google in Generative A.I?

Meta AI and Amazon are doing more than it looks like in 2023.

Author

Michael Kevin Spencer
May 21, 2023

Recently developments highlight how Generative A.I. are likely to be won by major Cloud players and digital Advertisers.

This article we’ll tackle Meta (Facebook) and Amazon.

Who do you believe will be most influential in the future of Generative A.I. and LLMs?

Login or Subscribe to participate in polls.

If you want to read the stories around the Web I’m reading on A.I. go to our Reddit.

  • For A.I. related jobs go here. If you are an A.I. company or have machine learning, software engineering or data science jobs you can post them here.

 

So let’s get into it:

Photo by Adrian Sulyok on Unsplash

AMAZON

If Generative A.I. and GPT-4 forces Google to remake our search experience online, the one company with the most to gain is not Microsoft, it’s actually Amazon.

Amazon has been taking marketshare in product search from Google for years and has become an advertising behemoth in its own right. I argue in this article that it will continue to do so and generously fund a new era of search in its own right.

Click on the title of this piece to read on the web or go here if you reach a Gmail limit.

Source: stratably

In early 2023, Amazon’s advertising business grew 19%, while Google and Meta both deal with slowdowns. (source)

📦 Amazon has a window, much more than BingAI, to reimagine its search interface, with the power of large language models, and it will. Too much is at stake for it not to do so.

While Google is innovating its future of search and A.I. products, and while Microsoft’s investment in OpenAI was a good decision, Amazon will spend as much as Meta A.I. in how Generative A.I. can super-charge its advertising growth, and has a better chance of reinventing the search interface.

📶 From our sponsor: 📶

Amazon Textract Innovation Series: Co-create the future of intelligent document processing with AWS and Mission Cloud!

AWS continues to innovate to help you drive higher business efficiency. Textract's recently released new features improve ease of use, accuracy and accelerate the time to get insights out of your documents. Discover best practices in Part 1 of our live webinar series on May 24th, 9am-10am PT.

Let’s get back to our topic.

Amazon already has AWS, if it can win more marketshare in digital Advertising, it will increase its diversification to take on robotics and other scalable ventures around logistics, the food industry and other lower margin big pieces of the puzzle. 🧩

LLMs and Generative A.I. will be a key way how Amazon does this. Not just with its Search but with Alexa.

Google and Meta are both aiming to use LLMs to give advertising new super-powers. However they must fend of newcomers ByteDance, Apple, Microsoft and others. Amazon by far has the fastest growth in North America, and thus has the most at stake in how they adopt LLMs to boost their advertising growth and diversification.

Rohit Prasad, Amazon’s senior vice president and head scientist for Alexa, said the e-retailer will be a major player in generative AI. Amazon said on Wednesday that more than 500 million Alexa-powered devices have been sold worldwide. Amazon therefore has the most to gain in building a viable and improved A.I. assistant and chatbot for consumers. It already has the existing network and search authority.

Since Amazon has the most to gain from Generative A.I’s impact on advertising, it needs to get it right. It will be in no rush to chase BingAI and Google’s own PaLM-2 innovations. Up until now, Amazon’s biggest entry into the market thus far has been through an AI service (CNBC) for cloud customers. However since ByteDance and Amazon have the most momentum in advertising growth, they will invest the most in LLMs that elevate their game.

It’s very naive to think Google and Microsoft as early adopters, will be the long-term winners here. Microsoft rushing a bunch of GPT-4 products or getting 75% of OpenAI’s revenue is more a diversification play than a hail mary. The Ads duopoly is crumbling, but when you have that level of dominance, it takes many years to play out.

Amazon’s Secret is Out

Amazon.com Inc. plans to bring ChatGPT-style product search to its web store, rivaling efforts by Microsoft Corp. and Google to weave generative artificial intelligence into their search engines.

ByteDance and Amazon will find the secret sauce between social commerce, chatbots, product search and E-commerce sales. All of this will be elevated with Generative A.I. In that respect, ByteDance and Amazon have the most to gain in search marketshare, not Bing or other poorer contenders.

Alexa could lean into more generalized large language models like the ones behind ChatGPT and Bard. And it will. Alexa is the most likely consumer device to arrive at a smart OS level of sophistication, inspite of its failures in real utility. LLMs could enable Alexa to be resurrected from the land of the dead.

Amazon ad revenue hitting $100 billion

Amazon’s advertising revenue will one day reach $100 billion. The advertising slowdown of the post pandemic era will eventually lead to a consumer recession. Generative A.I. will be the bounce-back in how the search interface and Ads are transformed by Generative A.I. in the mid to late 2020s.

The Ads duopoly of Google and Meta will be much more evenly distributed in the future between:

  • Google

  • Meta

  • Amazon

  • ByteDance

  • Apple

  • Microsoft

Just like Google joined the Cloud race, Amazon is maturing the fastest in the digital Ads cash-cow game. How it leverages LLMs to speed up its growth even faster is key to its future as a company and as an innovation leader (that has noticeably dropped off since Bezos departed as CEO).

Apple has a much higher ceiling in terms of advertising than say Microsoft, this is because of its B2C services walled garden. Amazon also has this specific advantage, shoppers already come to Amazon which is useful for hundreds of thousands of retail advertisers.

In early 2023, Amazon increased its advertising revenues from $31bn to $41bn, and many believe it will be able to reach $100 billion in the foreseeable future. The profitability of AWS enables Amazon to invest in becoming an advertising empire that is able to disrupt the duopoly.

New job postings spotted online reveal clues about the e-commerce giant's generative AI plans. According to new job listings posted by Amazon and spotted by Bloomberg (BNN), the company is hiring for different positions, including a senior software development engineer who would work on "reimagining Amazon Search with an interactive conversational experience".

It’s only common sense Amazon has to do this, but the job postings make for amusing evidence of how they are seeing this challenge. One job post read that it would be part of “a new AI-first initiative to re-architect and reinvent the way we do search through the use of extremely large scale next-generation deep learning techniques.”

While Microsoft has spent a lot on Bing, it never really stood a chance. Amazon on the other hand is a real B2B brand. It inevitably takes marketshare from the Duopoly after the iOS privacy change, after the Ads slowdown, and what TikTok is doing to both YouTube and Instagram.

However since LLMs will be a significant way digital advertisers improve their interface, make targeting more sticky and upgrade recommendations, Amazon has the most to gain from the sum-total of the innovation in the space we are going to see from 2023 to 2026 especially.

Amazon can Disrupt Consumer Search

The BingAI GPT-4 competitor to Google search was more for publicity for Microsoft pretending its an AI-native company. For Amazon, it’s not just for show. This is because platforms like Amazon and ByteDance have already been taking marketshare from Google search for years, for years!

Another job spotted by Bloomberg read:

Yes it will be a chance for Amazon to disrupt how consumer Search works for products. There will be even less point to go to Google. If ChatGPT takes marketshare in time spent on Google, it won’t impact Amazon’s search.

Since Amazon has smart technology products, it can also adopt generative A.I. into its devices, scale Alexa with new software, and create an interface that rises above Google in more ways. This means LLMs enable Amazon to become a diversified advertiser with even more moats.

  • Smart tech products in the home

  • Alexa

  • A better Search Interface

As impressive as PaLM-2 is, Google simply doesn’t have a great relationship with consumers. Meta’s products are stale after years of copying, cloning and mismanagement. I believe Amazon, Apple and ByteDance will disrupt the Ad Duopoly in the next five years, and LLMs will hasten the downfall.

BingAI is about as laughable to me as Cortana was in the smart speaker race, first to arrive and first to be forgotten. This isn’t our first rodeo in terms of consumer A.I. races. We know the players and history is the best predictor of future success. The big surprise of the 2020s will actually be Apple’s rise in advertising revenue. But that’s another story.

Yes Amazon is producing a conversational AI search function for shopping, but it will do a whole lot more. Amazon is best positioned to build Generative A.I. into robotics that are useful in the home. Amazon is best positioned to sell actually viable personal assistants for consumers in the home. Amazon is best positioned to elevate consumer product search for the global consumer with new immersive experiences.

Amazon’s response to Google Gemini will be significant. Google announced a lot of things at Google I/O earlier in May, 2023. Amazon just has to do a few things right with Generative A.I. in the 2020s to secure that it will continue to disrupt Google. Forget all the hyperbole of Microsoft even being in the picture, it’s not. It’s not even rational.

Amazon will Reimagine Advertising in the 2030s

Amazon’s job listing are just a confirmation, they aren’t a source. It’s clear for any strategist what Amazon has to do with Generative A.I. in the Cloud and in the Ads spectrum. It needs to reimaging Search and the search interface completely.

Here is a screenshot:

The other factor is ByteDance is just so much smarter when it comes to B2C consumer apps and products, and this is where Google is the weakest among the major BigTech companies.

Meta AI mostly innovated via acquiring Instagram and WhatsApp, it doesn’t have innovation in its DNA like ByteDance or to a lesser extent, Amazon. Product and scaling consumer apps experience is important in execution, not just great engineering. This means Google and Meta are weak from getting fat on Ad revenue for so long. You can tell by their leadership woes and internal chaos. For all their talent and riches, they aren’t well managed companies and it’s beginning to show.

These macro factors favor Amazon and Apple to take marketshare faster from Google and Meta than many expect or even see possible in the 2020s and 2030s. Generative A.I. and LLMs present yet another vehicle for this to occur. Amazon’s layoffs were pretty brutal when it comes to its hardware and Alexa divisions. With LLMs it needs a new strategy, and that will take time.

Amazon was always going to be looking into adding generative artificial intelligence (AI) to its search function. Amazon will learn from how Google integrates A.I. further into search, and it will improve upon it for its product search that’s more valuable to advertisers and brands.

With ChatGPT, an improved Amazon Search, TikTok getting better (if it’s not banned), Google will be getting less traffic. Even with substantial technical chops in A.I., Meta will struggle to keep its Ads empire entact. Amazon doesn’t even have to implement these things in a stellar way, but in the early 2030s, I predict it actually will. Amazon is the long-game winner of the future of ads, for obvious reasons.

It’s not Microsoft, but Amazon, and to lesser extent, Apple and ByteDance that changes the future in Search, advertising and building a better customer experience for people looking for products and experiences, and not just answers. And it’s so fricking obvious.

All of this before the non-invasive brain-consumer-interface (BCI) next iPhone moment for the consumer interface transition takes place. Guess who leads research in that area? It’s China.

If you’d like to Sponsor our growth and reach our valuable audience of A.I. enthusiasts, executives and software engineers, you are free to do so. I am accepting tasteful and aligned minimalistic native Ads with maximum top-of-the-email focus exposure for valuable clicks.

Amazon is just getting started in Ads

While the $38 billion that Amazon made from ads in 2022 is a small fraction of the company's overall revenue that year of $502 billion, ads "probably had well over 50% operating margins, which would mean it brought in as much operating profit as AWS," according to Benedict Evans.

  • In 2021 Amazon started splitting it out, and last year Amazon had $38bn of ad revenue, more than any traditional media company and bigger than the entire (now much reduced) global newspaper industry.

  • While aligning with pro open-source Generative A.I. startups like Stability A.I. and Hugging Face, Amazon is aligning with the side that’s really going to democratize LLMs. Who knows what Amazon Bedrock itself will become.

  • Generative A.I. doesn’t just repshape search, it reshapes the future of the Cloud and advertising.

On average, Amazon’s peers grew their ad business by 1.7% compared to Amazon's 21% Y/Y growth.

Actual 2023 growth

The A.I. arms race will enable the disruption of the ads duopoly to occur faster. I expect Amazon and ByteDance to razer focus on this. When infographics omit ByteDance (owner of TikTok) you omit an important part of the picture.

Artificial intelligence has already had a pervasive impact on our lives, with Generative A.I. it augments the Cloud and digital Ads in new ways. Amazon with AWS and fast growing ads business is obviously, one of the main beneficiaries of all of this.

Thanks for reading!

META AI

Photo by Dima Solomin on Unsplash

Meta A.I.’s contribution to open-source A.I. cannot be overstated in 2023, from LLaMA in March, Sernau, too, highlights Meta AI’s crucial role in his Google memo. In early May, Meta A.I announced ImageBind, a Holistic AI learning across six modalities. Meta says that its model, ImageBind, is the first to combine six types of data into a single embedding space.

Meta AI as signaled by their last earnings call, is going to be a focus on A.I. specifically moving into the second half of 2023 and 2024 for the company formerly known as Facebook. Mark Zuckerberg referenced AI about 27 times at Meta's earnings call in late April, and it sent the stock soaring. META 0.00 

I’m watching closely how A.I. will reshape how some of the most powerful companies in the world work growing new revenue streams within their business models and to boost their existing products and go-to profitability. LLMs are clearly going to play a big part in Amazon’s urgency to become a more dominant advertising player.

·

May 18

Hello Everyone, If Generative A.I. and GPT-4 forces Google to remake our search experience online, the one company with the most to gain is not Microsoft, it’s actually Amazon. Amazon has been taking marketshare in product search from Google for years and has become an advertising behemoth in its own right. I argue in t…

But as impressive as Meta AI’s contribution to this ‘Linux of A.I.’ moment in open-source, today I want to briefly cover their recent chip announcements. In the United States, more than 300 firms seek a slice of the $39 billion in US Chip funds. Recently Britain launched a $1.2 Billion semiconductor plan.

📶 From our sponsor: 📶

Amazon Textract Innovation Series: Co-create the future of intelligent document processing with AWS and Mission Cloud!

AWS continues to innovate to help you drive higher business efficiency. Textract's recently released new features improve ease of use, accuracy and accelerate the time to get insights out of your documents. Discover best practices in Part 1 of our live webinar series on May 24th, 9am-10am PT.

The Speed of A.I. Feels Irresistible in 2023

As much as I despise Facebook as an entity and what they have done to the world, their A.I. unit, Meta A.I., has substantial contributions to make to the community and to the rate at which A.I. progress will take place in the decade ahead. Those advertising dollars have to go somewhere.

Even as ChatGPT launches their mobile app, the “Speed of A.I.” does seem faster these days. Microsoft is also working on their own A.I. chips. The stock market also seems to think the speed of A.I. is peaking, where Nvidia NVDA 0.00 stock is up already 120% just this year so far, which is an absurd amount.

The best source for PR about Meta A.I’s innovations is actually their head Yann LeCun on LinkedIn.

MTIA v1: Meta’s first-generation AI inference accelerator

Covering hardware announcements is less flashy, but A.I. chips are becoming a big deal as LLMs are far more demanding and expensive.

Click on the title of this piece to read on the web or go here if you reach a Gmail limit.

Meta found that GPUs were not always optimal for running Meta’s specific recommendation workloads at the levels of efficiency required at our scale. So naturally their solution to this challenge was to design a family of recommendation-specific Meta Training and Inference Accelerator (MTIA) ASICs. So they co-designed the first-generation ASIC with next-generation recommendation model requirements in mind and integrated it into PyTorch to create a wholly optimized ranking system.

MTIA = Meta Training and Interference Accelerator ASICs.

A.I. Compute Systems are Incredibly Power Intensive

There are massive bottlenecks with LLMs at play.

What is MTIA v1?

Microsoft is reportedly working on its own AI chips that can be used to train large language models and avoid a costly reliance on Nvidia. Meta’s approach seems a bit more advanced.

Meta in 2020 designed their first-generation MTIA ASIC for Meta’s internal workloads. If you like specs it’s also fairly interesting. This inference accelerator is a part of a co-designed full-stack solution that includes silicon, PyTorch, and the recommendation models. The accelerator is fabricated in TSMC 7nm process and runs at 800 MHz, providing 102.4 TOPS at INT8 precision and 51.2 TFLOPS at FP16 precision. It has a thermal design power (TDP) of 25 W.

Reimagining Meta’s infrastructure for the AI age

What about A.I. focused data centers? Google, Microsoft, Amazon and Meta’s AI compute needs will grow dramatically over the next decade and they must all adapt to it, the “speed of A.I.” forces them down a path where they need to break new ground in AI research, ship more cutting-edge AI applications and experiences for their products. In this case of Meta, their “family of apps”, while continuing to build their long-term vision, things like the metaverse.

Specialized A.I. datacenters? They are in the process of building the next generation of Meta’s infrastructure backbone – specifically built for AI including their first custom chip for running AI models, a new AI-optimized data center design, and phase 2 of our 16,000 GPU supercomputer for AI research.

If you think about it, these are truly $Billion dollar moves few companies in the world can actually follow. All those Ad dollars Google and Meta made in digital advertising give them a huge advantage in leading the next era of A.I. They can also afford the premium talent of the researchers, engineers and product leaders to make these huge projects happen.

Watch the excellent video here.

Meta’s Research SuperCluster

Meta says at full strength, they achieve almost 5 exaflops of computing power. (One exaflop is a quintillion — that’s a billion billion — calculations per second.)

How these companies have to stack Nvidia hardware is really incredible. This level of performance is achieved through the use of 2,000 NVIDIA DGX A100 systems as RSC’s compute nodes, for a total of 16,000 NVIDIA A100 Tensor Core GPUs, connected via an NVIDIA Quantum InfiniBand 16 Tb/s fabric network. Okay then Meta!

They introduced their Meta AI Research Super Clusters (RSC) in January, of 2022. Suffice to say that Facebook has a huge advantage in Generative A.I. with their AI Research SuperCluster (RSC).

It’s literally contributing to the Speed of A.I. I keep talking about.

The speed of A.I. compounds with all these new variables taking place in 2023, that leads to a different trajectory to things like AGI and the automation of our jobs. I’m not sure a lot of humanity fully understands this quite yet. You reach a point where technological progress is no longer gradual. Even Andrew Ng doesn’t seem to realize this yet (at least not publicly): his recent interview with Bloomberg was fairly interesting.

My thesis is this “speed of A.I” especially in the 2025 to 2050 period is going to surprise some people.

The A.I. announcements of companies like Meta and Google are really pivotal in all of this.

A series of AI announcements by Meta:

  1. MTIA v1: an AI chip for fast inference: https://lnkd.in/eF3hfVcu

  2. RSC: 5 exaflops, 16,000 GPU Research Super Cluster for AI research: https://lnkd.in/e9F4jsbF

  3. AI-focused data centers: https://lnkd.in/eXPa3jhW

How China is approaching AGI is also worth watching. The entire A.I. arms race is starting to speed up due to the rivalry between the U.S. and China brewing on the geopolitical sidelines. A lot of government money is also being thrown at the problem, a lot of integrations about how supply-chains, corporations and powerful elite groups work together.

Artificial intelligence will be a key focus of the talks between members of the business and political elite in Lisbon, Portugal. All in, around 130 participants from 23 countries are set to attend the private meeting — a similar number to previous years. That’s a lot of centralization of decision making for something as important as the Speed of A.I.

Meta’s role in Open-source and the future of compute in artificial intelligence should not be underestimated. The legacy of Facebook has to amount to something. Over the past decade or so, Meta has spent billions of dollars recruiting top data scientists and building new kinds of AI, including AI that now powers the discovery engines, moderation filters and ad recommenders found throughout its apps and services. This spending won’t just continue, it will accelerate.

One last thing about the speed of A.I., Generative LLMs are going to accelerate innovation in robotics considerably. Thanks for reading!

This publication is seeking Sponsors

If you’d like to Sponsor our growth and reach our valuable audience of A.I. enthusiasts, executives and software engineers, you are free to do so. I am accepting tasteful and aligned minimalistic native Ads with maximum top-of-the-email focus exposure for valuable clicks.