The $600 billion Reality Check: Why AI is a lousy marketing manager

By Robert Nagy, General Manager of Product & Operations, Yango

If you believed the headlines over the last two years, we should all be out of a job by now.

We were told that generative AI was a paradigm-shifting business solution that would instantly overhaul how brands connect with consumers, how agencies operate, and how the entire marketing ecosystem functions. Instead, as we move through the back half of 2026, the industry is waking up to a massive financial and operational reality check.

The truth is that AI hasn't fundamentally transformed the core of marketing; it has just made certain niches slightly more efficient. As the tech hype cycle inevitably pivots to its next shiny object, it's time for brands and agency leaders to look at what AI actually is: an expensive, highly specific tool that still desperately requires human oversight.

The frontier model price tag

The biggest open secret in Silicon Valley right now is that running massive, "frontier" AI models is commercially unsustainable for everyday tasks.

According to analysis from Sequoia Capital, the global spend on AI infrastructure has reached a staggering $600 billion annually, yet the technology is failing to produce measurable revenue, or productivity gains proportional to that investment. Compute and energy costs are rising so rapidly that these massive models are increasingly locked behind expensive enterprise subscriptions and reserved for high-level, complex research projects.

For the average consumer and everyday marketer, the future isn't a supercomputer in your pocket. It is small language models (SLMs). These smaller, cheaper, task-specific models are quietly taking over everyday functions, including how search evolves. Traditional search engines as we know them are fading into ad-supported, AI-assisted discovery engines powered by SLMs, because it is the only way platforms can afford to keep them free and functional for the masses.

The marketing miscalculation and the "great rehire"

Perhaps the most damaging narrative of the past two years was that AI could seamlessly automate the entire brand-to-consumer relationship. Brands rushed to replace customer service teams, community managers, and creatives with algorithms, expecting massive cost savings.

Instead, they degraded their own customer experience and ended up paying a premium for it.

We are now seeing a massive course correction, often referred to as the "AI Boomerang." According to recent data from Robert Half, 34% of professional services companies that cut staff for AI have already been forced to reinstate those positions. Even worse, 75% of those employers found that the massive cost of implementing the AI, combined with the expense of rehiring and retraining human talent, completely erased their initial savings.

The corporate world is littered with recent examples of this false economy:

• Klarna famously froze hiring and replaced human customer service agents with an AI chatbot, only to watch customer satisfaction scores plummet. By mid-2025, their CEO publicly admitted that the AI lacked nuance and empathy, forcing the company to pivot back to human agents to protect brand loyalty.
• Locally, the Commonwealth Bank of Australia tried to replace dozens of service roles with an AI voice bot, but it ultimately failed to handle real-world ambiguity, driving up call volumes and putting excessive pressure on the remaining human support teams.
• Ford spent years replacing veteran engineers with automated systems, resulting in terrible product recall rates. This year, they quietly rehired hundreds of human experts because AI couldn't replicate the deep, undocumented judgment required to solve complex problems.

Brands are discovering that AI pricing models, which often charge "per token" or per query, can rapidly spiral out of control at scale. In many instances, maintaining software subscriptions, cloud infrastructure, and the human managers required to oversee the AI actually costs more than the original human workforce.

Controlled automation, not total transformation

Look under the hood of most enterprise AI success stories, and you won't find a transformative business overhaul. You will find highly specific, niche applications.

In marketing operations, AI is successfully filling process gaps: transcription, level-one customer service triage, basic code generation, and photo or video editing shortcuts. These efficiencies are undeniable, but they are incremental.

At Yango, our philosophy centres around "controlled automation". We use technology to automate the heavy lifting of data, media buying execution, and repetitive tasks, but we recognise its limits. In many operational scenarios, implementing a bespoke AI solution is actually more expensive - and less reliable - than simply utilising existing SaaS platforms or paying a human expert. AI is a fantastic operational assistant, but it is a terrible marketing manager.

If the last few years have taught us anything, it's that technology scales, but thinking does not.

As an industry, we need to stop viewing AI as a blanket replacement for talent and start treating it as a targeted operational tool. A brand can have the most advanced AI tech stack in the world, but if the strategic inputs are weak, the creative and commercial outputs will be generic.

At the end of the day, marketing is a people-focused, results-driven industry. We still rely on people to do the work that moves the needle. AI can process data, speed up editing, and format reports, but human ingenuity solves a client's business problem and connects with a consumer.

The tech industry will inevitably move on to the next big thing. The brands and agencies that win won't be the ones that replaced their teams with algorithms; they will be the ones that gave their best people the right tools to work a little smarter.