Nvidia on How AI Will Transform Adland in 2024

“One thing has become very clear to everyone; unlike some of the cycles of interest we've had in technology in the marketing space in the last few years, AI is by no means a flash in a hype cycle pan.”

Jamie Allan is the director of business development, global agencies and advertising at Nvidia, the tech company known for its hardware and software that has advanced the supercomputing, data science and gaming industries, among others. It has recently made headlines for reaching a $2 trillion market value (only the third American company in history), thanks to emerging as one of the world’s leaders in AI. 

Nvidia designs semiconductor chips used for generative AI tools, and provides foundational layers of hardware and software upon which the world’s largest enterprises are building their own platforms and products, AI-powered or not. Nvidia, and specifically Jamie, also work closely with partners across the advertising industry, from holding companies to production experts and even start-ups, assisting them with deploying AI, among Nvidia’s other expertise and offerings.

Speaking with LBB last year about unlocking the power of AI for advertisers, Jamie said the first half of 2023 had been “the ‘iPhone moment’ of AI”, sparking a year he’s retrospectively termed “the year of the pilot”. This year, however, he’s already deemed as “the year of platform and production”, telling LBB’s Ben Conway in a recent conversation that it’s “true table stakes” for every enterprise in the world. 

Above: Jamie Allan

Moving on From AI Pilots

“It needs to not be the product anymore,” says Jamie. “AI isn't ‘the thing’ - it’s how AI helps us create new things.”

Unlike other technology adoption curves, AI is moving very quickly, shown by the announcements at the start of the year, with the likes of WPP, Publicis and Media.Monks setting out their stall in the space with large monetary investments and clear strategies. “They set out an ethos of ‘we're building core, centralised platforms’,” he says. “Publicis with CoreAI and WPP building into Open with its ‘brand brain’ strategy and production studio strategies… and there's examples in other places too: Omnicom’s Omni as well as efforts within IPG, Dentsu, and all the others.”

This centralisation of AI is vital, as Jamie explains that it being siloed or used on the edges for tests and pilot projects will make it difficult if a company wants to keep up. “No one's going to have the amount of investment or intelligence to build something really meaningful that impacts your whole company.”

In the agency landscape specifically, he shares that leaders also have an opportunity to become trusted advisors in the AI space, guiding brands on their own journeys - something he’s seen companies at the forefront of the change double-down on. With AI and its associated change mechanisms being complex to understand, from a technology and business practice point of view, he says that understanding it - not just for yourself, but for your clients - is the first big step. 

“[The industry is] moving beyond the exciting AI pilots we saw last year, moving towards making it a fundamental element of the products and services that they need to offer.” 

However, a lot of this leads to considerations around business models. Brands’ expectations around AI are changing, and as efficiency and cost savings enter the conversation, many industries will need to evaluate how they position and cost a service or product that is more automated.

“The quicker the business models can be adapted to the impact of AI, the more successful companies will be as well,” adds Jamie. “That's something agencies have the opportunity to help guide brands on, once they become experts in that business transformation. That's a really interesting part of the evolution of what an agency can mean to a brand; rather than just being about their campaigns and marketing, being about their future direction.”

A Tricky Transformation

While integrating and centralising AI can be difficult for any size of company, deploying new ideas and paradigms about how a business operates, and even values its work, is a fundamental challenge that may favour leaner, more agile set-ups.

“If I'm looking at these technologies and tools today as a mid-sized organisation, it's a huge opportunity to be more dynamic and innovative than perhaps a larger organisation could, because of the scale that they would need to change at,” explains Jamie.

“AI models and these computing capabilities are hugely accessible. The democratisation of AI, from an academic practice to an API call, makes it incredibly easy to build out a product or service. We've seen this with the explosion of startups in the last 12 months. The same could be said for those smaller agencies, media companies and production studios.”

Describing it as “a blessing and a curse”, he notes that smaller companies have an enhanced ability to pivot and make the business - although larger organisations have the scope to assemble teams that can find solutions very quickly also. “It's exciting at both ends of the spectrum. I love working with smaller companies and startups. Nvidia traditionally puts a huge amount of emphasis in that area. Our startup ecosystem in our Inception programme has over 20,000 members now, and I'm regularly speaking to new startups in this space.”

Above: Nvidia Omniverse Avatar Cloud Engine

In the world of production - where Jamie expects to see significant progress in AI involvement throughout 2024 - he says that there’s a huge volume of incredibly excited and motivated leaders who see the new tools and automation processes as ‘supercharging’ what they do. “It’s what they've actually dreamt of doing for some time,” he says. 

“We haven't not wanted to bring mass personalisation to advertising, right? We haven't not wanted to connect the world of 3D data and artificial intelligence and campaigns before. Now the tools are available for people to execute on what the visionaries in the industry have been talking about for some time.” 

However, the challenges arise from building enterprise platforms stable enough to achieve that. “Lots of amazing pilots, innovation labs and tests have happened in the last 12 months - but how do we scale this? That’s a challenge, but not impossible to overcome in a short amount of time, as long as you collaborate between creatives and production leaders who know how that pipeline needs to work, and IT professionals, AI engineers and cloud DevOps teams who know how to make that work.”

Traditionally, these haven’t been super connected groups; they haven't really needed to be. But Jamie believes bringing those things together can solve the challenge - preparing companies for the impact of transforming their creative processes, and their business itself, across the next 12 months.

Combining AI and Data: The “Next Big Step”

After a year of pilots and tentative AI activity, Jamies says that 2024 will also see many agencies ask the question: “How do you connect the power of generative AI with the power of data?”

“And that's going to be the next big step this year,” he says.

Publicis has spoken on this with its Core AI platform and how it connects data from Epsilon with its capabilities across Sapien, Publicis Media and its production teams. WPP has discussed collaboration between Group M, Satalia and Hogarth. Media.Monks announced its Flow platform at CES. All looking at things in slightly different ways, says Jamie, “but fundamentally understanding that AI requires data, intelligence and creative production (at scale) to have a big impact.” 

“Generative AI especially is not about the generation of content, but the generation of intelligence. And the quality of that intelligence is based on the data, the sources and the teams building those models and pipelines.” He continues, “If you are generating and owning data, then you should own the intelligence that that data is going to produce as well. And you should have the capability to generate that intelligence. There's a very good analogy of this: if we are milling the flour and selling it, and then buying back the bread, we aren't in the best position.”

Understanding the data you’re using to create or fine tune AI models and processes is very important, so having your own proprietary data is inevitably a huge advantage. For example, Nvidia has worked with Getty Images to create a foundational AI model for image generation based solely on the media company’s content, making the pair first movers in the space of licensable AI-generated imagery. Nvidia has also followed that with companies like ServiceNow in the IT enterprise services industry, and Amdocs in the telecommunications and media industry, combining their existing data, historical capability and knowledge bases to create proprietary tools. 

“If you consider that in the lens of large agencies, media companies and holding companies, yes, a huge amount of proprietary data about markets, audiences and campaign history is going to enable them to create very powerful AI models”.

Above: Nvidia's Maxine AI Video Streaming Platform

Though, that’s not to say the data is everything. Just having the data alone isn’t really enough to create efficient, high-quality AI models. You need AI engineering, data scientists, MLOps, management platforms and all sorts of things that Nvidia and its hyper-scaler partners like Meta and Google have been building for the last 5-10 years.

“There is a lot more to it than just the proprietary data,” says Jamie. “There are organisations that don't have a huge amount of data, but are building very advanced AI capabilities and AI teams to take their clients’ proprietary data and help them to create models, processes and tools for themselves. And we also have examples of the inverse, where there are large data companies still figuring out the ‘how’, ‘where’ and ‘who with’.” 

“If you have a huge amount of data, then building an AI model or fine tuning a model for a specific task can be relatively expensive, so you need to understand what the best use of that data is going to be,” he adds. “But fundamentally, having large amounts of data that's commercially safe for you to use, that you can extract more value from - whether it's with language model fine tuning or image/video generation fine tuning, and connecting them together - will certainly be of huge value.”

Should Creatives Be Worried?

Discussions around AI in adland - including in this very article - are so often centred around ‘efficiency’ - new paradigms, processes and business transformations largely in the name of doing more with less. It’s not unreasonable to suggest then, that the advertising industry’s creatives and creators - anyone along the pipeline who will be increasingly interacting with AI - might have some concerns about what ‘efficiency’ and ‘transformation’ means for them, in real terms. 

Jamie suggests that companies “need to remain” both creative and efficient throughout these transformations. “When you have a clear direction on how these tools are going to impact your work and adapt to them, and embrace that, it’s an exciting thing.”

“Jobs will be augmented and supercharged, especially in the creative side,” he continues. “The best in the industry are looking at these tools and setting out very flexible strategies about their creative pipelines - how they can integrate multiple tools and not be set in a single creative process.”

To create work in a new way where it’s easy to adopt and test tools, Jamie says that companies should empower their creatives and maintain a level of enterprise mindset, strengthening innovation, R&D and a wide array of other teams to ensure a consistent production and AI capability.

“It's about skilling up legal, skilling up compliance, skilling up ethics - all these teams in the organisation. Last year, we saw a lot of that starting to happen, allowing the landscape of agencies to quickly understand the capabilities of a tool, from a creative or an AI engineering point of view, and also how likely it is that a particular model, tool or service will fit into their production layer, rather than exclusively their R&D and innovation layer.” 

He adds, “Looking at AI purely as ‘it's going to make us more efficient’, in isolation, is a very hard task. You need to apply the thinking of ‘how this will improve and supercharge all of our work’ to bring efficiency.”

Above: Nvidia's Gr00t Humanoid

2024: The Year of Platform and Production

“If we move from ‘the year of pilot’ in 2023, we start to look at ‘the year of platform and production’,” says Jamie. This evolution from technological testing and pilot productions - companies dipping their toes into the AI pool - into an era of decisive AI integration and centralisation will see larger companies building out their platforms to become true enterprise software and service platforms.

“And across individual agencies and smaller agencies, [we’ll see] how they move into production with these technologies,” he adds. 

From a consumer lens, Jamie anticipates that the general population will begin to experience this new AI-powered stage of marketing through the much-anticipated arrival of what he refers to as “unintrusive personalisation”.

“It’s an interesting shift,” he says, “the idea of personalisation-at-scale, from content production, and using proprietary data to create privacy-first personalisation that can bring an era of a more attractive, dynamic one-to-one advertising. Many years of research have shown that it can drive better brand engagement and growth, and improve return on ad spend.”

This is already available, he says, to “the Googles and Metas of the world”, who have been building and developing platforms and new tools like Advantage Plus and Emacs which enable this and allow for scaling. “Those solutions and tools will ultimately drive that unintrusive personalization.” 

So whether it's more unique customisation in the production pipeline, greater creative capabilities, business efficiencies, or new uses yet to be innovated, AI is seemingly going to play an ever-increasing role in adland. And, to stay at the cutting edge, Jamie suggests that, in 2024, it’s time to stop testing the waters, and instead, prepare to jump in; transforming your business to centralise AI before the opportunity sails downstream.

“What we're seeing right now is the worst AI is ever going to be. So, you build your ideas today around something that's going to constantly improve.”