From Hype To Hyperscale In AI
Recorded: Dec. 2, 2025, 12:03 p.m.
| Original | Summarized |
From Hype To Hyperscale In AI | AdExchanger
image/svg+xml:
Topics Opinion About Us Events Podcasts Programmatic I/O New York Become an AdHero Subscribe Sign In
Sign In Topics
Home AdExchanger Talks From Hype To Hyperscale In AI
PODCAST: AdExchanger Talks Tuesday, December 2nd, 2025 – 6:00 am
It’s like watching a movie, he says. How do you know if it’s good before you even try it? Online advertising is evolving from a reliance on carefully engineered features and audience segments to harnessing the power of these “foundational models” that can learn from huge volumes of raw data, Ahn says. Subscribe AdExchanger Daily Daily Roundup Daily News Roundup While generative AI models are currently pretty good at answering questions, most are way too slow and expensive to meet the needs of real-time advertising, he says, which demands more powerful models that can operate instantly and affordably.
Next In AdExchanger Talks Making Your Brand Matter To The Models
Related Stories PODCAST: AdExchanger Talks PODCAST: AdExchanger Talks PODCAST: AdExchanger Talks PODCAST: AdExchanger Talks Must Read Technology AudienceMix, a new curation startup, aims to make it more cost effective to mix and match different audience segments using only the data brands need to execute their campaigns. Digital Out-Of-Home On Tuesday, digital out-of-home (DOOH) ad tech startup Place Exchange was acquired by Broadsign, another out-of-home SSP. Social Media For the uninitiated, “Glitchmas” is our name for what’s become an annual tradition when, from between roughly late October through November, Meta’s ad platform just seems to go bonkers. antitrust The publisher-focused DOJ v. Google ad tech antitrust trial is finished. A judge will now decide the fate of Google’s sell-side ad tech business. Programmatic Competitive tensions and ad tech drama have flared all year. And this drama has rippled out into the investor circle, as evident from a slew of recent ad tech company earnings reports. Platforms Two sources at ad tech platforms that observe programmatic bidding patterns said they’ve seen Omnicom agencies shifting spend from The Trade Desk to Amazon DSP in Q3. The Trade Desk denies any such shift.
Popular OPINION: Data-Driven Thinking CTV spending is flattening, performance is plateauing and buyers are hesitant to push budgets further. The reason is not complicated. When buyers cannot see what they are buying, they cannot commit their spend with conviction. Marketers Manscaped’s CMO Marcelo Kertész on flipping its strategy from quick-win performance marketing to building a lasting brand with smarter, long-term growth. Social Media For the uninitiated, “Glitchmas” is our name for what’s become an annual tradition when, from between roughly late October through November, Meta’s ad platform just seems to go bonkers. Digital Out-Of-Home On Tuesday, digital out-of-home (DOOH) ad tech startup Place Exchange was acquired by Broadsign, another out-of-home SSP. PODCAST: AdExchanger Talks You can’t buy your way to the top of a large-language model. At least not yet. But there are things that brands can do to influence how – and if – they get mentioned, says Tracy Morrissey, SVP of media and performance at full-service agency Innocean USA.
Join the AdExchanger Community Your trusted source for in-depth programmatic news, views, education and events. NEXT EVENT
ABOUT ADEXCHANGER CONNECT
© 2025 Access Intelligence, LLC - All Rights Reserved |
This AdExchanger Talk podcast transcript, featuring Ikkjin Ahn of Moloco, offers a valuable perspective on the evolving landscape of artificial intelligence within the digital advertising industry. The core message centers around the shift from theoretical AI “magic” to the demonstrable power of “AI hyperscale,” emphasizing the critical need for substantial computational resources to effectively harness the potential of large language models. Ahn correctly identifies a key differentiator – the ability to process massive volumes of data – as a crucial element for separating genuine progress from marketing hype. The discussion highlights Moloco’s use of Google’s AI Hypercomputer, illustrating a practical application of this concept and highlighting the costs associated with attempting to meet real-time advertising demands with slower, less scalable approaches. The podcast touches upon the limitations of current models, emphasizing the need for faster, more affordable solutions that can operate in real-time, while also briefly addressing topics such as brand influence on large language models and broader trends in the ad tech ecosystem, including the acquisition of Place Exchange. Reflecting the broader AdExchanger audience, the transcript incorporates references to ongoing industry debates, such as transparency concerns within CTV, and coverage of related news items including Omnicom’s shift in spending, and Meta’s perennial holiday platform glitches. |