Speed kills coordination — and right now, the teams moving fastest on AI adoption are experiencing the sharpest communication breakdown precisely because the tools designed to help them move faster are also making it easier to skip the conversations that hold teams together.

Dave Rupert's recent essay lands like a quiet alarm bell in a noisy room. The title says it plainly: "When moving fast, talking is the first thing to break." It's a deceptively simple observation, but the implications run deep — especially when you layer in what's happening simultaneously across the AI chip wars, the consumption economy, and the way AI-generated text is now seeping into professional communication in ways that are subtly eroding the signal quality of human interaction.

Why This Moment Is Different From Every Previous "Move Fast" Era

We've heard the "move fast and break things" mantra for over a decade. But the current inflection point is structurally different in at least two ways.

First, the speed tools available today — LLMs, agentic AI, code assistants — don't just accelerate output. They replace the intermediate steps that used to force communication. When a developer had to ask a colleague how to implement something, that friction generated alignment. When a product manager had to write a proper spec because the team couldn't read their mind, that friction generated shared context. AI tools increasingly allow individuals to skip those friction points entirely, producing outputs that look complete but carry none of the embedded organizational knowledge that the old friction-based process generated.

Second, the economic pressure to move fast is intensifying in ways that compound the problem. The "Great American Contraction" narrative — where traditional labor is being systematically repriced as AI capabilities expand — means that teams are smaller, timelines are shorter, and the margin for the kind of slow, deliberate communication that builds real alignment is shrinking. When headcount is down and output expectations are up, "let's schedule a sync" becomes a luxury that gets cut.

The irony is that the very tools meant to reduce cognitive load are increasing the hidden coordination debt that eventually surfaces as misalignment, rework, or — in the worst cases — team fracture.

The AI Text Problem: When the Medium Betrays the Message

Here's a data point worth sitting with: Le Monde recently ran a piece on how ChatGPT usage is being identified in professional writing — specifically through the telltale use of em-dashes and particular sentence constructions. This isn't a trivial stylistic observation. It points to something structurally important about what happens to communication quality when AI-generated text starts flowing through organizational channels.

When a colleague sends you a Slack message or an email that was substantially drafted by an LLM, several things happen simultaneously:

1. The signal-to-noise ratio of their actual thinking drops. You're reading a polished output, not a window into their reasoning process.
2. The social contract of communication changes. Professional writing has always carried implicit signals — urgency, confusion, enthusiasm, uncertainty — that help recipients calibrate their response. AI-smoothed text strips those signals.
3. The feedback loop breaks. If someone is confused about a project direction but uses AI to write a confident-sounding status update, the confusion never surfaces. The manager reads a clean summary and assumes alignment. The misalignment compounds silently.

This is the communication breakdown that Rupert is pointing at, even if he's approaching it from a slightly different angle. The problem isn't just that teams talk less when they move fast. It's that the quality of the communication that does happen is degrading in ways that are hard to detect until something breaks badly.

The Jensen Huang Dimension: What Pressure Does to Communication

Nvidia CEO Jensen Huang's recent heated exchange during a podcast with Dwarkesh Patel — where he reportedly "nearly lost his composure" when pressed on the China chip question, declaring "You're not talking to someone who woke up a loser" — is instructive here, though perhaps not in the way the tech press framed it.

What that moment illustrates is what happens to communication under extreme strategic pressure. Huang is navigating one of the most complex geopolitical-commercial tensions in the history of the semiconductor industry. The pressure to move fast (capture market share, maintain revenue, satisfy shareholders) is in direct conflict with the pressure to communicate carefully (navigate export controls, manage government relationships, maintain public trust).

His near-composure-loss isn't a personal failing — it's a diagnostic signal. When the stakes are high and the pace is relentless, even the most experienced communicators start to show cracks. The organizational version of this plays out in every startup, every enterprise AI team, every product organization that's trying to ship faster than its communication infrastructure can support.

The Nvidia situation also highlights a broader geopolitical communication breakdown that has direct market implications. The US-China AI chip dynamic is being shaped by a series of miscommunications, misaligned incentives, and policy signals that are interpreted differently by different actors. The result is a market environment where companies like Nvidia are simultaneously being told to move fast (by shareholders and customers) and slow down (by regulators and government officials) — and the communication infrastructure for navigating that tension is visibly strained.

The Organizational Debt Nobody Is Measuring

Here's the frame I keep coming back to: communication debt is the new technical debt, and it's accumulating faster than most organizations realize.

Technical debt is the accumulated cost of shortcuts taken during development — it's measurable, trackable, and most engineering organizations have at least some framework for managing it. Communication debt — the accumulated cost of conversations not had, context not shared, alignment not established — is far harder to measure and almost never tracked.

When teams move fast with AI tools, they're often trading communication debt for speed. The problem is that unlike technical debt, which tends to surface gradually through slower development cycles and increasing bug rates, communication debt tends to surface suddenly and catastrophically — in the form of a product launch that misses the market, a team that fractures under pressure, or a strategic pivot that nobody actually agreed on.

The consumption collapse narrative is relevant here too. As economic pressure mounts and organizations are forced to do more with less, the temptation to skip the "expensive" communication steps — the all-hands meetings, the strategy offsites, the 1:1s that feel like they could have been emails — becomes overwhelming. But those are precisely the investments that prevent the sudden, catastrophic surfacing of communication debt.

The Cloud Governance Parallel

This dynamic maps directly onto something I've been tracking in the enterprise AI adoption space. The same pattern that creates communication breakdown in teams — moving fast, skipping the governance steps, accumulating invisible debt — is playing out in how organizations are deploying AI in their cloud infrastructure.

As I explored in my analysis of AI tools and cloud trust decisions, AI tools are now making consequential decisions about cloud access and trust that used to require explicit human sign-off. The governance conversation — who authorized this? what are the guardrails? who's accountable when something goes wrong? — is being skipped in the rush to ship. And as I noted in the analysis of what gets logged in AI-driven cloud environments, the logging and attribution infrastructure that would allow organizations to reconstruct those skipped conversations after the fact is often not in place.

The parallel is not coincidental. It reflects a deeper pattern: when you move fast with AI, the first thing to break is the communication layer — whether that's human-to-human communication within a team, or the governance communication between an AI system and the humans who are supposed to be overseeing it.

What "More Talk, Less Grok" Actually Means in Practice

Rupert's prescription — more talk, less relying on AI to synthesize and summarize your way to understanding — is right, but it needs to be operationalized carefully to be useful. Here are the specific communication investments that appear to have the highest return in fast-moving, AI-augmented teams:

1. Protect the "Confusion Surfacing" Moments

The most valuable communication in any organization is the moment when someone says "I don't understand" or "I'm not sure we're aligned on this." AI tools are very good at making it easy to avoid those moments — to produce a confident-sounding output that papers over genuine uncertainty. Teams need to actively create spaces where confusion is safe to surface, and where AI-polished communication is explicitly set aside in favor of raw, unmediated human exchange.

Weekly "confusion dumps" — unstructured time where team members can surface things they don't understand without fear of looking uninformed — are one mechanism. Requiring that major decisions include a "dissent log" where anyone can record disagreement or uncertainty is another.

2. Audit Your Communication for AI Smoothing

If your team's Slack channels, email threads, and status updates are increasingly polished and confident-sounding, that's not necessarily a sign of improving communication quality. It may be a sign that AI smoothing is hiding the signal. Periodically audit the actual information density of your team's communication — are people sharing genuine uncertainty, or are they sharing AI-summarized certainty?

3. Invest in Synchronous Communication During Ambiguous Phases

Research on distributed teams consistently shows that synchronous communication — real-time conversation, whether in-person or video — is significantly more effective than asynchronous communication for resolving ambiguity. (MIT Sloan Management Review has published extensively on this.) The temptation when moving fast is to default to async because it feels more efficient. But for the genuinely ambiguous, high-stakes decisions, that efficiency is illusory — you're paying for it later in rework and misalignment.

4. Make the Communication Infrastructure Visible

Most organizations have detailed visibility into their code infrastructure (version control, CI/CD pipelines, deployment logs) but almost no visibility into their communication infrastructure. Who talked to whom about what, when? What decisions were made in which conversations? What context was shared and what was assumed? Building even a lightweight communication log — decision records, context documents, explicit "here's what we discussed and agreed" summaries — creates the audit trail that allows organizations to reconstruct the conversations that didn't happen when they needed to.

The Bigger Pattern: Speed as a Communication Strategy

There's a deeper game theory dynamic worth naming here. In competitive markets, moving fast is often a deliberate strategy for preventing communication — specifically, for preventing competitors from having time to respond, for preventing internal dissenters from having time to organize, for preventing regulators from having time to react.

This is the "move fast and break things" strategy at its most intentional. The problem is that it works on external actors and internal dissenters, but it also works on the communication infrastructure that the organization itself needs to function. You can't selectively suppress communication — when you create an environment where speed trumps talk, you suppress the conversations you want to suppress and the conversations you desperately need to have.

The organizations that appear to navigate this best are those that have learned to distinguish between the communication they can afford to skip (routine status updates, process documentation, low-stakes coordination) and the communication they cannot afford to skip (strategic alignment, risk surfacing, genuine dissent). The former can be handled by AI tools and async processes. The latter requires protected time, psychological safety, and a genuine commitment to slowing down enough to actually hear what people are saying.

---

The signal in Rupert's essay is worth amplifying: the communication breakdown that comes with moving fast isn't a side effect that can be patched later. It's a structural consequence of the way AI tools are changing the economics of communication — making it cheaper and easier to produce polished outputs while making it more expensive and more uncomfortable to surface genuine uncertainty and disagreement.

The teams and organizations that figure out how to protect their communication infrastructure while still moving fast will have a durable competitive advantage. The ones that don't will keep shipping fast until the accumulated communication debt surfaces all at once — and by then, the speed that felt like a strength will have become the source of the fracture.

More talk. Less grok. The advice is simple. The discipline to follow it, in an environment that rewards speed and punishes friction, is anything but.