AI Has Moved from the Lab to the Balance Sheet
Good morning,
AI just made its way from the lab to the balance sheet.
This week, Meta is building new models (Mango and Avocado), OpenAI is reportedly chasing an $830B valuation, and Pew Research shows that over half of Americans now interact with AI multiple times a week. But behind the headlines, something bigger is shifting.
In Big Picture, we break down the move from AI experiments to AI infrastructure, and what it means for CMOs under pressure. Engagement is shifting from search to answers. Investors are rewarding defensible stacks over optionality. And CFOs aren’t funding AI as a narrative, they’re funding what proves value.
Also in this issue:
Why reasoning models just hit a usage tipping point
What’s really behind the CMO turnover surge
And why the fastest-growing AI startups in 2026 will look more like ERP vendors than prompt tools
The hype cycle is over. The operating cycle has begun.
- The Marketing Embeddings Team
NEWS
Meta is rolling the dice again in the AI arena, this time with code names straight out of a fruit bowl. The company is developing two major generative AI models: Mango, focused on image and video, and Avocado, a large language model aimed at improving text and code capabilities. (Read more)
OpenAI is reportedly planning to raise as much as US$100 billion in fresh capital, potentially pushing its valuation to a staggering US$830 billion. (Read more)
According to a Pew Research Center survey, 27% of Americans reported interacting with AI at least several times a day, while another 28% said they interact with AI about once a day or several times a week. (Read more)
AI AND THE CMO REVOLVING DOOR
This week’s corner is intentionally a step back from the technical details.
Instead of models, prompts, or platforms, it looks at the people who ultimately decide whether marketing AI succeeds or fails inside an organization: CMOs. Two recent Adweek articles, one examining the biggest CMO shakeups of 2025 and another spotlighting leaders to watch in 2026, offer a revealing lens into where marketing is heading, and why AI adoption is no longer just a technology discussion.
The first signal is volatility. The CMO role has become one of the most unstable positions in the C-suite. But this isn’t churn for churn’s sake. It reflects a fundamental redefinition of the role. Today’s CMO is increasingly expected to act as a Chief Transformation Officer, responsible not only for brand and growth, but for reshaping how marketing operates across data, systems, and culture. AI sits squarely at the center of that expectation.
At the same time, CMOs are facing a “power and proof” reckoning. CEOs and CFOs are pushing marketing to move away from soft indicators like awareness and toward clear commercial impact. ROI has become the primary budget metric, and every major initiative must now justify itself in financial terms. In this environment, AI that feels experimental is easy to deprioritize; AI that proves value is not.
Another shift emerging from these profiles is how AI itself is being positioned. In 2025, AI was often treated as an experiment at the edges. Heading into 2026, it is increasingly viewed as infrastructure—the operating system for marketing. That includes changes like moving from SEO to generative engine optimization, and from manual execution to agent-driven workflows that compress time and reduce friction.
Yet CMOs face a critical constraint: fragmentation. Most are surrounded by data but lack insight, speed, and coherence. Managing a stack of disconnected tools makes it nearly impossible to move quickly or tie activity back to revenue.
The implication for marketing AI builders is straightforward. CMOs don’t need more tools. They need leverage, clarity, and fast wins. Solutions that reduce complexity, connect directly to financial outcomes, and deliver value quickly are far more compelling than platforms that promise transformation “over time.”
In an era of constant turnover, the most successful AI will be the kind that helps a CMO survive, and win, their first 90 days.
BIG PICTURE
From AI Experiments to AI Infrastructure
Marketing, AI, and capital markets are converging around a single, unavoidable truth: experimentation is over, and proof now matters more than promise.
According to the Podcast from the AI Daily Brief about 51 charts explaining AI in 2026. The winners in 2026 will not be the loudest AI storytellers, but the operators who can turn AI into measurable business outcomes—fast.
1. Attention Is Shifting from Search to Answers—and Engagement Follows
The first chart highlights a quiet but profound shift: traffic from generative AI sources like ChatGPT is behaving fundamentally differently than traditional search traffic.
Non-bounced visitors referred by ChatGPT spend 3× more time on site, view more pages, and convert at higher rates than Google referrals. This is not just a traffic story—it’s an intent story. Users arriving from generative answers are already pre-qualified. They are not browsing; they are continuing a decision journey.
For CMOs under pressure to defend spend to CFOs, this matters enormously. It signals the transition from SEO to GEO (Generative Engine Optimization) and reinforces why marketing leaders are being recast as Chief Transformation Officers. Owning AI-native distribution is no longer a brand play—it is a revenue lever.
2. Markets Are Repricing AI—from Optionality to Defensiveness
The divergence between OpenAI-exposed and Alphabet-exposed stocks tells the same story from the capital markets side.
Investors are moving away from pure model speculation and toward industrial AI stacks—those with control over compute, predictable capex, and cash-flow visibility. Alphabet increasingly represents AI defensiveness; OpenAI represents AI optionality. And right now, markets are choosing defensiveness.
This mirrors what CMOs are experiencing internally. CEOs and CFOs are no longer funding AI as a narrative. They are funding it as infrastructure—or not at all. AI initiatives that cannot demonstrate operational leverage, cost reduction, or revenue impact are being deprioritized.
For AI startups, this is a warning and an opportunity: build like infrastructure, not like a feature.
3. Reasoning AI Has Crossed the Rubicon
The OpenRouter data confirms that reasoning models now account for roughly 50% of total usage, a tipping point that changes how AI must be designed, deployed, and governed.
This is critical because reasoning models are not toys. They are slower, more expensive, and vastly more capable—making them suitable for agentic workflows, not one-off prompts. This aligns directly with what enterprise leaders are demanding: fewer tools, more outcomes.
It also explains the renewed interest in roles like a Chief Data, Analytics, and AI Officer, as organizations struggle with fragmented data, decentralized experimentation, and AI systems that are powerful but disconnected from daily work.
What This Means for AI Builders and Buyers
Taken together, these signals converge on a single mandate:
AI must operationalize, not inspire
Speed to value beats feature depth
ROI visibility is the new moat
CMOs are exhausted, CFOs are skeptical, and CEOs want proof. The AI startups that win in 2026 will not be those that add yet another dashboard, model, or plugin—but those that collapse workflows end-to-end, bridge marketing outcomes to financial metrics, and deliver measurable impact in weeks, not quarters.
In short, AI has officially moved from the lab to the balance sheet. And the market, both human and financial, is already voting.
