If the classic narrative arc is one of rise (hype), fall (disappointment) and redemption (ROI) then AI is starting, in some eyes, to fall into that part of the film where the protagonist does stupid things and their friends leave.
“A CIO canceled a Microsoft AI deal. The reason should worry the entire tech industry” as one Business Insider headline put it this week, citing lacklustre results from the expensive Copilot. (Most developers The Stack speak to say it tends to generate bloated boilerplate and is of little use helping to generate new business logic. If CTOs listen to both their developers and their CFO closely, expect more such headlines…)
So far, so “trough of disillusionment” in some quarters.
Over at Meta, meanwhile, Mark Zuckerberg has executed an energetic pivot away from the “metaverse” (arguably never likely to make it to the “redemption” phase) and to generative AI, where Meta has become a major influence with its regular open-source Llama model releases.
(Let access to 16,000 NVIDIA H100 GPUs and a $40 billion annual Capex allocation be your friend if you can – that’s $100 million every day.)
"Not a meaningful driver of revenue"
Executives freely admitted on a Q2 earnings call on July 31 that ROI on generative AI was still elusive; expect questions to continue mounting.
Meta CFO Susan Li said that despite this colossal spending “we don't expect our Gen AI products to be a meaningful driver of revenue in 2024.
“We do expect that they are going to open up new revenue opportunities over time that will enable us to generate a solid return [on] investment while we are also open sourcing subsequent generations of Llama.”
That, however, will also not be cheap.
“The amount of compute needed to train Llama 4 will likely be almost 10 times more than what we used to train Llama 3, and future models will continue to grow beyond that. It's hard to predict how this trend – how this will trend multiple generations out into the future. But at this point, I'd rather risk building capacity before it is needed rather than too late, given the long lead times for spinning up new inference projects.”
That was Mark Zuckerberg, speaking on the same call.
Meta, of course, has money to experiment with. Total Q2 revenue was $39.1 billion, up 22%. Net income, meanwhile, was a healthy $13.5 billion
Meta AI Studio: Ghosts in the shell
“Traditional” AI is helping drive more engagement across Meta’s family of applications (Facebook, Instagram, Threads, WhatsApp et al) which have a combined total of over 3.5 billion active daily users, Meta noted this week.
This week (in the US, to start with) Meta also spun up its “AI Studio” that lets users create AI avatars, no doubt with an eye to further engagement.
“You can use a wide variety of prompt templates or start from scratch to make an AI that teaches you how to cook, helps you with your Instagram captions, generates memes to make your friends laugh – the possibilities are endless,” Meta said of its AI studio. (A world in which AI characters laugh mechanically at each others’ memes and teach each other to cook pretend food in a digital echo chamber hermetically sealed off from the planet their creators are baking to facilitate this? We’re long on that play.)
In May, Meta also began rolling out full image generation capabilities into Advantage+ Creative advertising platform, and CFO Susan Li said “we're already seeing improved performance from advertisers using the tool.”
To Zuckerberg, the long-term ad revenue opportunities from genAI are compellingly substantial: “It used to be that advertisers came to us with a specific audience they wanted to reach, like a certain age group, geography, or interests. Eventually, we got to the point where our ad systems could better predict who would be interested than the advertisers could themselves… advertisers still need to develop creative themselves.
“In the coming years, AI will be able to generate creative for advertisers as well. And we'll also be able to personalize it as people see it,” he added.
“Over the long term, advertisers will basically just be able to tell us a business objective and a budget, and we're going to go do the rest…”
Meta’s deep reach and AI chops will also improve its business messaging proposition, Zuck suggested, as enterprises aim to move away from often frustratingly limited chatbots and trim what are frequently huge support overheads: “Business AIs are the other big piece here,” he said on the call.
“We're still in Alpha testing with more and more businesses. The feedback we're getting is positive so far. Over time, I think that just like every business has a website, a social media presence, and an email address, in the future I think that every business is also going to have an AI agent that their customers can interact with. And our goal is to make it easy for every small business, eventually every business, to pull all of their content and catalog into an AI agent that drives sales and saves them money.
“When this is working at scale, I think that this is going to dramatically accelerate our business messaging revenue,” Zuckerberg added.
CFO Susan Li hastened to add that “Our ongoing investment in core AI capacity is informed by the strong returns we've seen and expect to deliver in the future, as we advance the relevance of recommended content and ads on our platform. While we expect the returns from Generative AI to come in over a longer period of time, we’re mapping these investments against the significant monetization opportunities that we expect to be unlocked across customized ad creative, business messaging, a leading AI assistant and organic content generation.
“As we scale generative AI training capacity to advance our foundation models, we’ll continue to build our infrastructure in a way that provides us with flexibility in how we use it… This will allow us to direct training capacity to gen AI inference or to our core ranking and recommendation work, when we expect that doing so would be more valuable.”
It’s clearly still the “rise” phase at Meta. Will there be a “fall”?
Share your thoughts on generative AI ROI and production use case value.
