EXECUTIVE BRIEF
This week's intelligence reveals a critical acceleration in the AI-AdTech landscape, demanding immediate executive attention. The era of programmatic advertising as we know it is ending, replaced by a new stack where AI-native platforms, owned infrastructure, and algorithmic superiority define market winners. The key takeaway is that value is shifting from traditional intermediaries to vertically integrated players who control their own technology and data from the ground up. The business implication is clear: companies that fail to adapt to this new, AI-driven operating model will face margin compression and rapid loss of competitive advantage.
Prioritised Recommended Actions:
Priority | Action | Impact | Effort |
|---|---|---|---|
1 | Pilot the Agentic RTB Framework | High | Medium |
2 | Test AppLovin's Self-Serve Platform | High | Low |
3 | Cease Foundational Model R&D | High | Low |
STRATEGIC ANALYSIS
1. The Agentic Shift: IAB Tech Lab Unveils AI-Powered Programmatic Framework
TL;DR: The IAB Tech Lab has introduced the Agentic RTB Framework, a new technical standard that uses containerised AI agents to revolutionise programmatic advertising. This move dramatically cuts latency and shifts strategic power from platforms to advertisers and publishers, enabling real-time, sophisticated decision-making directly within the bidstream.
What Happened: The IAB Tech Lab released the Agentic RTB Framework for public comment, with a planned Q1 2026 rollout. This new standard moves programmatic auctions into a localised, containerised environment, allowing AI agents from various parties (advertisers, publishers, data providers) to interact in real-time. This reduces auction latency from 400-600 milliseconds to a fraction of that time [1]. Key players like Index Exchange, Amazon Ads, Publicis, and WPP are already involved in shaping the framework.
Why It Matters: This is a foundational shift in the architecture of programmatic advertising. It dismantles the current, fragmented system of slow, sequential API calls and replaces it with a hyper-efficient, collaborative environment. The competitive landscape will now favor those who can deploy the most effective AI agents, giving advertisers unprecedented control over their bidding logic and allowing publishers to enrich their inventory on the fly. This move commoditises the role of traditional DSPs and creates a new battleground based on algorithmic superiority.
Competitive Opportunities:
Gain First-Mover Advantage: Develop and deploy proprietary bidding agents to outperform competitors who are still reliant on slower, less sophisticated DSP-based bidding.
Unlock New Data Collaborations: Leverage the framework's secure container environment to partner with data-rich entities (e.g., retailers, financial institutions) for real-time audience enrichment without direct data exposure.
Enhance Publisher Value: Publishers can integrate real-time fraud detection and data enrichment agents to increase the value of their inventory and command higher CPMs.
Executive Actions:
Action | Owner | Timeframe | Expected Outcome |
|---|---|---|---|
1. Launch Pilot Program: | CTO / Head of Programmatic | Q1 2026 | Assess performance gains and cost savings of the Agentic RTB Framework with a key campaign. |
2. Form Strategic Alliances: | Chief Strategy Officer | Q4 2025 | Engage with IAB Tech Lab and framework participants (e.g., Index Exchange, Chalice) to influence standards and secure early access. |
3. Develop In-House Agent Expertise: | Head of Data Science | Next 6 Months | Build or acquire talent to create proprietary bidding agents that align with specific business goals, creating a durable competitive moat. |
Key Data Points & Sources:
The new framework can shrink a transaction that used to take 400-600 milliseconds down to a fraction of that time [1].
Early adopter Chalice is already seeing 50% week-over-week growth with its self-serve platform built on similar principles [2].
The framework enables secure collaboration with data that has been on the sidelines, such as show-level CTV data or bank purchase data [1].
Strategic Framework Mapping: Go-to-Market
"The future of programmatic isn't about speed, it's about folding time. The Agentic RTB Framework puts AI in the driver's seat."
2. AppLovin's AI Engine Delivers 68% Growth, Signals AdTech Power Shift
TL;DR: AI-native ad platform AppLovin reported a stunning 68% year-over-year revenue surge to $1.41 billion in Q3 2025, driven by its Axon 2.0 AI engine. This performance, contrasted with the struggles of traditional DSPs like The Trade Desk, signals a major market shift where integrated, AI-first platforms are winning investor confidence and market share.
What Happened: AppLovin's Q3 2025 earnings showcased explosive growth, with revenue hitting $1.41 billion and adjusted EBITDA soaring 79% to $1.16 billion [2]. The company's stock is up over 80% year-to-date, while competitor The Trade Desk has seen its stock fall by 63% in the same period. The growth is attributed to the success of its Axon 2.0 AI platform and the recent launch of a self-serve platform that is already seeing 50% week-over-week growth.
Why It Matters: AppLovin's success is a clear indicator that the ad tech market is bifurcating. On one side are legacy DSPs built for the cookie-based open web, which are now facing headwinds. On the other are AI-native, vertically integrated platforms like AppLovin that combine ad inventory, bidding intelligence, and now, creative generation. This model offers superior performance and is rapidly expanding from its core in gaming to broader categories like retail, creating a significant competitive threat to the established order.
Competitive Opportunities:
Capture Non-Gaming Verticals: Leverage AppLovin's new self-serve platform to target non-gaming audiences (e.g., e-commerce, finance) with high-performance campaigns before competitors catch on.
Optimize Ad Creative with AI: Utilize AppLovin's planned AI-driven ad generation tools to create more effective, platform-specific creative, boosting ROAS.
Reallocate Ad Spend for Higher ROI: Shift budgets from underperforming traditional DSPs to AI-native platforms that demonstrate superior, data-backed results.
Executive Actions:
Action | Owner | Timeframe | Expected Outcome |
|---|---|---|---|
1. Test Self-Serve Platform: | CMO / Head of Growth | Q4 2025 | Allocate 10% of a major campaign budget to AppLovin's new self-serve platform to benchmark performance against incumbent DSPs. |
2. Establish AI Creative Taskforce: | Chief Creative Officer | Next 3 Months | Form a small team to pilot AI-driven ad creative generation, preparing for the rollout of AppLovin's new tools. |
3. Conduct Portfolio Review: | CFO / CMO | Q1 2026 | Review ad tech vendor performance and reallocate at least 20% of budget from the bottom quartile of performers to top-quartile AI-native platforms. |
Key Data Points & Sources:
AppLovin's Q3 2025 revenue surged 68% to $1.41 billion, with adjusted EBITDA up 79% to $1.16 billion [2].
The company's new self-serve platform, launched October 1, 2025, is experiencing 50% week-over-week growth [2].
AppLovin's stock is up over 80% YTD, while The Trade Desk's stock has declined by 63% in 2025 [2].
Strategic Framework Mapping: Monetization
"While others talk about AI, AppLovin is shipping it. A 68% revenue surge proves the market rewards performance, not just promises."
3. The AI Profitability Crisis: HBR Warns of a Broken Business Model
TL;DR: A new analysis from Harvard Business Review highlights a fundamental flaw in the generative AI business model: high variable costs and low variable revenue. The report argues that foundation model builders like OpenAI are unlikely to be profitable, while the real value is captured by infrastructure providers (like Nvidia) and established platforms (like Meta) that use AI as a feature, not a product.
What Happened: Harvard Business School professor Andy Wu, writing in HBR, argues that generative AI companies face a structural profitability challenge due to the high variable "inference" costs incurred with every user query [3]. With OpenAI expecting to spend over $150 billion on these costs by 2030, the current $20/month subscription model is unsustainable. The analysis concludes that foundation models are becoming a commodity, and the companies building them are in a poor position to capture long-term value.
Why It Matters: This is a critical strategic insight for any business investing in AI. The hype around building proprietary foundation models is misplaced and capital-intensive. The analysis suggests that the most viable strategy is not to build the core AI, but to be a smart user of it. Companies with existing distribution, data, and user bases (i.e., moats) are best positioned to win by integrating commoditized AI to enhance their existing offerings. This fundamentally reframes the "build vs. buy" debate, steering smart money away from foundational R&D and toward application-layer innovation.
Analyst Inference: The AI industry is experiencing a 'value-capture inversion.' The companies absorbing the most risk and capital expenditure (the model builders) are on a path to becoming low-margin utilities. The highest-margin opportunities lie in the application layer, where AI can be deployed to solve specific business problems within an existing, defensible business model.
Competitive Opportunities:
Focus on Application, Not Foundation: Instead of building large language models, focus on creating high-value, vertical-specific AI applications using existing APIs (from OpenAI, Anthropic, etc.).
Strengthen Existing Moats with AI: Integrate AI features into existing products and platforms to increase user retention, value, and competitive differentiation.
Avoid the AI Arms Race: Resist the temptation to enter the capital-intensive race to build the largest model; instead, treat foundation models as a commodity input and focus on business model innovation.
Executive Actions:
Action | Owner | Timeframe | Expected Outcome |
|---|---|---|---|
1. Cease Foundational Model R&D: | CTO | Immediate | Halt any internal projects focused on building general-purpose foundation models and redirect resources to the application layer. |
2. Identify AI Integration Opportunities: | Chief Product Officer | Next 3 Months | Map the top 3 customer pain points where existing AI APIs can be integrated to deliver immediate value and strengthen the product's moat. |
3. Model for Usage-Based Costs: | CFO | Q1 2026 | Revise financial models to account for the variable, usage-based costs of AI, ensuring product pricing scales with AI consumption. |
Key Data Points & Sources:
OpenAI expects to spend over $150 billion on inference costs through 2030 [3].
The variable cost of generating a single AI image is several cents in electricity and chip capacity [3].
The biggest stock market winner in the AI boom has been Nvidia (the "shovel seller"), followed by Meta (the "jewelry maker"), not a pure-play AI company [3].
Strategic Framework Mapping: Moat
"The AI gold rush is a myth. The real money is in selling the shovels (Nvidia) or making the jewelry (Meta), not in digging for gold (OpenAI)."
4. Anthropic's $50B Bet on Owned Infrastructure Signals AI's Next Chapter
TL;DR: Anthropic is investing $50 billion in a partnership with neocloud provider Fluidstack to build custom data centers, signaling a strategic shift from renting cloud capacity to owning the AI stack. This move highlights that for frontier AI companies, controlling the entire infrastructure stack is becoming a critical competitive necessity.
What Happened: Anthropic announced a $50 billion partnership with U.K.-based Fluidstack to build custom AI data centers in Texas and New York, set to come online throughout 2026 [4]. This is Anthropic's first major move to build its own infrastructure, a departure from its reliance on cloud partners like Google and Amazon. The move aims to maximize efficiency for Anthropic's specific workloads as it scales toward its projected $70 billion in revenue by 2028.
Why It Matters: The era of AI companies being purely software-driven is ending. As models become more complex and compute-intensive, owning the underlying infrastructure is becoming a key differentiator. This vertical integration allows for workload optimization, cost control, and supply chain security that cloud providers cannot match. It also signals the rise of "neocloud" providers like Fluidstack, which are emerging as credible alternatives to the AWS/Google/Azure oligopoly for specialized AI workloads. This trend will force every AI-dependent company to re-evaluate its infrastructure strategy.
Competitive Opportunities:
Explore Neocloud Providers: Evaluate emerging neocloud providers for specialized AI workloads to potentially achieve better performance and lower costs than traditional cloud giants.
Optimize Cloud Spend: Conduct a thorough audit of current cloud expenditures to identify workloads that could be run more efficiently on alternative or custom infrastructure.
Secure Compute Capacity: As the AI infrastructure arms race intensifies, secure long-term compute agreements to avoid being priced out or left without capacity.
Executive Actions:
Action | Owner | Timeframe | Expected Outcome |
|---|---|---|---|
1. Conduct Infrastructure Strategy Review: | CTO / CIO | Next 6 Months | Analyze the long-term cost and performance trade-offs of renting from hyperscalers versus partnering with neocloud providers or building custom infrastructure. |
2. Initiate a Neocloud Pilot: | Head of Engineering | Q2 2026 | Run a pilot project with a leading neocloud provider to benchmark performance and cost for a specific AI workload against current cloud solutions. |
3. Model Total Cost of AI Ownership: | CFO | Q1 2026 | Develop a financial model that projects the total cost of ownership for AI initiatives, including both training and inference costs across different infrastructure options. |
Key Data Points & Sources:
Anthropic is investing $50 billion in custom data centers, projecting $70 billion in revenue by 2028 [4].
The project marks a major success for neocloud provider Fluidstack, which is also the primary partner for an $11 billion AI project backed by the French government [4].
This move is part of a larger trend, with Meta committing $600 billion and the Stargate partnership (SoftBank/OpenAI/Oracle) planning $500 billion in infrastructure spending [4].
Strategic Framework Mapping: Ops Efficiency
"The AI battle is moving from the cloud to the ground. Owning your infrastructure is the new competitive moat."
STRATEGIC SYNTHESIS
The four breakthrough stories analyzed this week converge on two powerful, interconnected themes that are reshaping the AI and AdTech industries: The Great Value Inversion and The Rise of the Vertically Integrated AI Stack.
First, the Great Value Inversion describes a fundamental shift in where profits are made. As highlighted by the HBR analysis on AI's profitability crisis, the immense capital being poured into building foundational AI models is not where value is being captured. Instead, profits are flowing to the "shovel sellers" who provide the underlying infrastructure (like Nvidia, and increasingly, custom data center players like Fluidstack) and the "jewelry makers" who use AI to enhance existing, defensible platforms (like Meta and AppLovin). For business leaders, this means the strategy of building a proprietary large language model is a siren song leading to a low-margin, commoditized future. The real opportunity lies in the application layer, where AI can be wielded as a tool to solve specific business problems and strengthen existing moats.
Second, this value inversion is fueling the Rise of the Vertically Integrated AI Stack. From Anthropic's $50 billion investment in custom data centers to AppLovin's AI-powered, closed-loop ad platform, the most forward-looking companies are taking control of their entire technology stack. The IAB's new Agentic RTB Framework is another manifestation of this trend, disintermediating traditional players and allowing advertisers to bring their own intelligence directly into the bidstream. The era of relying on a fragmented ecosystem of cloud providers, DSPs, and other intermediaries is over. Winning in the new landscape requires owning or deeply controlling the infrastructure, the data, and the intelligence layer to create a seamless, efficient, and defensible system.
Prioritized Roadmap for a Mid-Size Digital Ad Business:
Q1 2026: Foundational Realignment.
Immediately halt any R&D on proprietary foundation models. Reallocate data science talent to focus on building proprietary bidding agents for the new Agentic RTB Framework.
Allocate a test budget (10-15%) to AI-native platforms like AppLovin to benchmark performance against your current DSP stack, focusing on non-gaming clients.
Begin a strategic review of your infrastructure, modeling the total cost of ownership for AI workloads and exploring neocloud provider partnerships.
Q2 2026: Offensive Maneuvers.
Launch a pilot program on the Agentic RTB Framework with a key client, leveraging your new in-house bidding agents.
Based on Q1 test results, create a transition plan to shift a significant portion of ad spend (30-50%) to top-performing AI-native platforms.
Formalize a partnership with a neocloud provider for a specific, high-cost AI workload to test performance and cost benefits.
Q3 2026: Scale & Differentiate.
Scale your use of the Agentic RTB Framework across multiple clients, marketing your proprietary bidding agents as a key competitive differentiator.
Develop a new service offering around AI-driven creative optimization, leveraging the capabilities of platforms like AppLovin.
Q4 2026: Strategic Moat-Building.
Use the savings and performance gains from your new AI-driven operations to fund a strategic acquisition of a complementary data or analytics company.
Publish a forward-looking whitepaper on your success with the new AI stack, cementing your firm's position as a leader in the next generation of advertising technology.
References: [1] Digiday - WTF is the Agentic RTB Framework? [2] The Motley Fool - AppLovin: Is It Time to Buy the Stock as Revenue Continues to Surge? [3] Harvard Business Review - AI Companies Don't Have a Profitable Business Model. Does That Matter? [4] TechCrunch - Anthropic announces $50 billion data center plan