NAB 2025: The AI Imperative
At NAB 2025, the term AI reigned supreme over other still-popular tech buzzwords like SaaS, cloud, and end-to-end that dominated previous’ years shows. But before jumping or falling into the AI world, perhaps learning from the past is a good idea. As with SaaS and cloud, buying or launching a family of products with AI does not make a strategy.
Over the years, the media industry has faced many challenges. Foremost is the disruption of prevailing business models. Then there’s “If we build it, will they come?” Can subscription, ad-supported, or hybrid publishing be profitable? How do we slow churn? Or compete with social? Or get more ad dollars? Merge linear and digital capabilities? Monetize our libraries?
Then there have been major tech transformations required: CapEx to OpEx, moving to HD, IP and multi-platform delivery. Several years ago, I heard Josh Steinhour, Principal Analyst, Devoncroft Partners say the problem with the media technology side of the house is every media company is a special snowflake and creates their own distinct tech stack, which in turn means vendors cannot bring a very standardized product to market and this makes it hard to grow or become very profitable. The upshot is we have companies building instead of buying off the shelf. Hold this thought.
At the NAB 2025 Devoncroft Executive Summit, attendees gathered the day before the trade show opens to hear thought leaders discuss where they see things going. Steinhour posed the question to media companies, “If you were late with file-based [asset deliver] or moving to IP, what was really the punishment for that? It probably wasn’t existential.”
When it comes to late adoption or sluggish adaptation, AI is different.
“We can make some guesses about where we are with AI,” Steinhour said, regarding the current state of AI adoption. “Our habit here is, whatever you build today, you can’t possibly touch again or reimagine or think about it again for about five years. You poured the concrete, right? That tends to be the approach in the industry, and especially with reference to early adopters. That would suggest that the decisions that are being made over the next 12 months are going to disproportionately impact the media industry for the next decade.”
I truly hope he’s joking—or at least exaggerating for effect.
“Sensitivities are high, but in serious times, all who would be called serious, have to have serious discussions,” Steinhour said. He showed a slide comparing operating margin profiles between 2015 and 2025. It did not look good. AMC went from 27% to -2%, the Walt Disney Company from 28% to 17%, Globo from 22% to 11%, CBS 17%/Viacom 23% (then) to Paramount (now) -18%, Discovery 31%/Warner Media 24% then to Warner Bros. Discovery -26%.
Repeating History?
Many companies are still working on moving to the cloud. The Canadian Broadcasting Corporation (CBC) sent out a tender to understand what could replace traditional hardware. 12 vendors qualified, 170 media functions were identified for a use case of delivering 80,000 hours of live content. “They also asked those vendors which platform they were supporting today and in the future. None of the vendors were willing to support any of their competitors,” said Steinhour.
“There's a real interoperability question there, but there are larger business goals we shouldn’t lose sight of,” he said. “Their primary consideration for their live production technology stack was to have the fastest path to arrive at the business goal of lowest price per hour production costs.”
Then there’s social. “YouTube disclosed within Alphabet, their annual advertising revenue is $36 billion,” said Steinhour. “We don't have a crisp number for subscription revenue, but we know there's 125 million subscribers. With YouTube music and YouTube premium it's a number north of $10 billion."
Of Google, Steinhour went on to say, “The parent company of YouTube does not see fit to source third-party technology for YouTube. The revenue-sharing arrangement with the creators that are posting to YouTube, that could be as much as 55% of that pie of revenue. There’s a lot of media companies that are now building that out as a distribution outlet.”
If trying to get vendors to work together to virtualize live playout was hard, or even get consumers to show up on your owned and operated streaming service, what will you do when every part of the media workflow can be disrupted by a bunch of smart AI startups?
AI Prerequisites
How do you start using AI? Do you start with internal media company teams and see what they come up with? Do you go to vendors? These were among the questions Steinhour raised. In another session one executive from a large media company mentioned he was considering building their own LLM. I was sitting with an executive from AWS when we heard this. We smiled. That might be taking things a bit too far.
“I do think it’s important to start having a conversation about how is the work that’s being done by media going to change? We continue to apply an old paradigm to thinking about how media workflows should move forward," said Anil Jain, Global Managing Director, Strategic Consumer Industries, Google Cloud. “These [AI] capabilities are now unlocking huge potential, which means you don’t have to do things the way you used to and when you couple that with what’s happening in the marketplace with creators, with other forms of content, with immersive experiences, the entire landscape of media will change.”
Jain continued, “Take a moment and rethink ‘Should our software work the same way it’s always worked?’ Is this the way we should be creating, producing, distributing, and monetizing content?”
“Every single one of the people in this room have different requirements about how to work with this technology,” said Simon Crownshaw, Worldwide Strategy Director Media & Entertainment, Microsoft. “There are use cases in this room that people want to solve that are all very different, but some of them have very general applications that are very similar: localization, subbing, dubbing, playout, upscaling, and so on.
“There is no question that data is the most important thing, whether you’re using it for customer data or content data, understanding how to use that data for with existing models. For the longest time, we haven’t taken as good care of the data as we should. What we see on many other customer engagements, let’s look at the data. Is it ready to be applied against the AI models I’m about to use?”
“There are almost 2,000 [AI] models that are sitting in the foundry. It is reasonable for people to say, ‘Every two weeks there's some completely different model. I’m supposed to engage with some completely different concept.’ How do you even organize around that?" said Steinhour.
“There is no one model to rule them all.... You have to be open to what the new possibilities are. You should not think about, ‘Do I have to build this from the ground up?’" said Jain. “I think that that is probably the wrong approach. Over the last couple of years, we can say that a little more definitively, there are capabilities out there, the costs are coming down, the agility and the speed at which you can actually make changes.”
Jain went on to describe prior surveys with media executives and recalled that “it was something like 83% of the 260 media executives that were surveyed globally said that they had taken less than six months to actually move a gen AI project from experiment to production.”
“Have we really gotten our head around the cost model for these and how much of that is a conversation that's coming up?” Steinhour said. “How much of these are recurring activities that are incurred future costs? I mean, some statements were made about the need to re-index on a time-to-time basis. Where is cost in this discussion?”
Jain responded, “For years, we have talked about OpEx and CapEx and moving from an on-prem to a cloud model or a hybrid model. I would challenge all of us to think, ‘Is the cost model and the economic structure of the way we’ve operated media businesses going to be the same moving forward?’ Of course, you can’t predict where the costs are going. We just know that. as [happened] with cloud costs, [in] the AI domain, performance is increasing, and costs are coming down, but you’re probably also going to scale up your usage.”
“We love media workflows,” said Crownshaw. “They live forever, and we wrap them with duct tape and all the rest of it, and they keep going. But ultimately, if the business model is kind of broken, then we need to look differently about how we build that going forward because ultimately that's just going to be critical to the bottom line,” he continued. “If the future of media content is going to be more digital, on every platform, then I need to be able to do multi-platform publishing really fast because information has never changed to this extent. It’s not just a question of ‘How do I get content from A to B?’ It’s ‘How do I get that content into multiple formats, into multiple platforms with the right information, the right insights at the right time, automatically?’’”
“Maybe we need to be a little more provocative here,” Jain asserted. “We come to NAB or IBC, and we see a number of the same vendors year after year after year. There are a number of vendors that are now on the market that are showing up that aren’t in this room today. The capabilities, because of the democratization of Generative AI combined with the ability to do things much more quickly, spin up new capabilities on the cloud, and to leverage data in different ways and to not be encumbered by legacy in terms of the way we’ve done things in the past (and what we have to continue to support), means that there’s a number of upstarts here [at NAB]. I think that we should be paying attention to them and I’m sure all the vendors here are paying attention to them as well.”
My Take
Many of the established vendors I spoke with at NAB did announce AI-powered services, and when asked, a good number of them said they had capabilities but not customers. Although I’m neither buying nor selling technology, I have concerns. However, when I spoke with a number of startups and innovators who had great approaches, I found an amazing breadth of product development, including:
- The Austrian Broadcasting Corporation (ORF) developed an application they license out to other media companies as well as make available for their employees of their 4 TV and 12 radio stations to scrape other media websites (and translate to German), summarize, and identify questions for their journalists to use to develop further content.
- UIC Digital had a great reference app they created to demo consumers interacting with custom AI-created sports clips.
- Suitest showed me their no-code testing for OTT apps and devices which parses code and OCR to put streaming apps through thorough testing.
- I attended a press conference where Haivision, Nvidia, Verizon, and FanDuel TV showed off local use of AI in proxy feeds where a chatbot interface can request which of the N+ cameras should be used to target and stream a specific high-res version using limited spectrum to deliver.
- I saw an AWS demo to create consistent character design and storyboards based on poster image and text input.
The problem and opportunity is, AI will change things. People will lose their jobs, processes will be sped up, repetitive tasks will be automated, and consumers expectations will change. I don't want to take away from the downsides of this and there are certainly a number of things. However, the media industry has been especially slow to endorse change, require a lot of custom development, and even keep up with the Joneses (namely Netflix and YouTube).
There is a whole very interesting world of AI-powered applications and models coming into the market. And perhaps remembering this is a different kind of tech. In addition to getting the use case right, AI can be very compute- and prompt-intensive. Figuring out how things can both work in your existing workflow, be backwards-compatible and interoperable is super-important. There is so much more and I’m hoping that if AI is resetting workflow requirements that anyone using it does much more due diligence this time around. That is unless you still want to hear the argument from established vendors that none of them are willing to support any of their competitors, as they swiftly go out of business.
Related Articles
In this NAB 2025 interview with Streaming Media contributing editor Jan Ozer, Lily Bond, Vice President of Marketing at 3Play Media, discusses how the company has evolved from one of the earliest captioning providers into a full-service accessibility and localization partner. With 15 years in the business, 3Play now offers captioning, subtitling, audio description, and hybrid AI/human dubbing services for customers looking to comply with accessibility laws and expand into global markets.
15 Apr 2025
In this interview with Streaming Media contributing editor Jan Ozer, Narayanan "Raj" Rajan, CEO of Media Excel, reflects on his first year at the helm of one of the most established names in the encoding industry. Rajan discusses how Media Excel is reasserting its relevance with the Hero platform—an encoder/transcoder/packager suite—and why live streaming, low latency, and AI-optimized encoding remain at the heart of the company's strategy.
15 Apr 2025
It's surely no surprise that a lot of companies at NAB are talking about including AI or Gen AI in the latest product releases. The challenge is to find companies whose AI implementations are driving the product vs. those that are just window dressing. Invariably, with some the proportion of AI hype to substance is too much, too little, or just right.
10 Apr 2025
On April 7 at NAB 2025's Streaming Summit, conference chair Dan Rayburn and LG Ad Solutions CMO Tony Marlow squared off in a fireside chat on "Performance Storytelling in the CTV Era." From the outset, Marlow declared, the challenge and the goal LG Ad Solutions sets for itself is to "monetize CTV premium ads from the moment you turn on your TV."
08 Apr 2025
AI-driven dubbing has recently gained attention as major platforms like Amazon Prime Video and YouTube roll out new tools designed to expand their content's global reach. Amazon is testing AI-assisted dubbing on licensed content, while YouTube has introduced auto-dubbing for thousands of channels. Both efforts reflect a growing belief that dubbing can help platforms engage new audiences—but the results so far have been mixed.
19 Mar 2025
For Anil Jain, leading the Strategic Consumer Industries team for Google Cloud has meant helping traditional media companies embrace cloud and AI, move to an OpEx-centric business model, and adapt to a more development-oriented mindset in engineering and product management. In this interview, Jain discusses the global shift to cloud-based operations in the media industry, the ways generative AI is disrupting everything from production to packaging to all aspects of the user experience, and what media companies should be afraid of if they're not already.
12 Jun 2024
Incorporating generative AI into our workflows has the potential to impact almost everything in media technologies, and I'll examine several of the possibilities in this article.
04 Oct 2023
Companies and Suppliers Mentioned