Big Tech's Sales Pitch: Do as I Do
Meta's Mark Zuckerberg is building an AI agent to help him run the company. Jensen Huang says Nvidia's engineers should spend $250,000 a year on AI tokens. Salesforce's Marc Benioff talks constantly about "digital workers" as a coming budget line item.
These aren't casual observations. They're a coordinated sales strategy.
Tech executives are using themselves as proof of concept. They're not just telling customers AI works - they're saying they use it themselves, heavily. It's the Hair Club for Men approach: the CEO is the testimonial.
What matters is how neatly this messaging aligns with what these companies sell. At Meta, aggressive internal AI adoption now shows up in performance reviews. At Nvidia, productivity gets measured by token consumption. At Salesforce, the future org chart runs on "digital labor."
The dogfooding play
Tech companies have long used their own products internally to improve them. But AI adoption feels different. Most people buying AI don't fully understand it yet. They're being told it's inevitable while watching the people selling it claim to depend on it.
The companies push internal adoption first, then point to that adoption as proof it works. The uncomfortable part: they might be both early and self-interested simultaneously.
AI probably does increase individual productivity. Agents probably will change work. Compute probably will become essential infrastructure, like cloud spending before it. The question is whether the urgency being broadcast matches the evidence.
The data gap
Business spending on AI has grown, and contract sizes have expanded, according to Ramp, a corporate expense platform. But external evidence of AI productivity gains remains thin.
Federal Reserve data shows 40% of adults use AI at work, yet time saved amounts to only 2% of total work hours. A Resume.org survey of 1,000 hiring managers found AI's impact minimal: 9% said it had fully replaced roles, 45% said it had little to no impact on staffing. One large real-world study found AI boosted customer service productivity by 15% on average, with gains concentrated among less experienced workers.
In the absence of solid proof, marketing fills the void.
How norms become strategy
Companies don't just buy software - they copy norms. When Nvidia's CEO says serious engineers use massive amounts of compute, that message spreads beyond Nvidia. Other companies begin evaluating their own teams by the same standard.
When Zuckerberg describes a flatter, agent-driven organization, it stops being a Meta experiment. It becomes a managerial benchmark other executives feel pressure to match.
Each company benefits when the definition of competence expands demand for what it sells. Nvidia wins when engineers use more tokens. Salesforce wins when every company believes it needs AI employees. Meta wins when agents run constantly.
The real pressure
There's a simpler explanation for the push: AI infrastructure is expensive. The biggest tech companies are spending hundreds of billions on data centers, chips, and power. That fixed cost only works if usage climbs steadily.
The message shifts from "this might help" to "you should already be doing much more of this." The faster AI becomes essential, the faster those massive investments look justified.
It's an effective way to sell a very expensive product to people still figuring out what it's for.
For executives evaluating AI strategy, the lesson is straightforward: distinguish between what works and what benefits the people selling it. The two aren't always the same.
Learn more about AI for Executives & Strategy to develop a clearer framework for these decisions.
Your membership also unlocks:
