a16z just argued that AI won’t destroy software companies because traditional moats — network effects, brand, proprietary data, process power — still hold. They’re right about most software categories. They’re wrong about marketing. Marketing software moats aren’t just weakening — they’re becoming liabilities. The same switching costs that kept customers locked in are now keeping incumbents locked out of the AI-native future.

Key Takeaway: Hamilton Helmer’s Seven Powers framework shows why most software survives AI disruption. But applied specifically to marketing software, it reveals something the a16z analysis misses: every moat that protects HubSpot, Salesforce Marketing Cloud, and the MarTech stack is a structural barrier to becoming the AI agent that replaces the marketer’s labor — not just their tools.

Alex Immerman and Santiago Rodriguez at a16z just published “Good News: AI Will Eat Application Software” — an optimistic case that software’s competitive advantages survive the AI transition. Their analysis is rigorous. They apply Hamilton Helmer’s Seven Powers framework systematically and conclude that while AI creates “a great software bifurcation,” the moated winners keep winning.

We agree — for most categories. But marketing software is different. Here’s why.

The Seven Powers, Applied to Marketing Software

Helmer’s framework identifies seven sources of durable competitive advantage: switching costs, network effects, scale economies, brand, cornered resources, process power, and counterpositioning. a16z argues most of these protect incumbents from AI disruption.

Let’s test that claim against marketing software specifically.

Switching Costs: The Moat That Became a Prison

a16z acknowledges this is the moat most weakened by AI. They write that AI “is changing the friction and the cost-benefit analysis associated with switching vendors” and that companies with high switching costs may have “hostages, not customers.”

For marketing software, this is already playing out. HubSpot’s moat was never the software — it was the pain of migrating CRM data, email templates, workflow automations, and reporting dashboards. Teams stayed because leaving was expensive.

But AI agents don’t need to migrate your data out of HubSpot. They need to make HubSpot irrelevant. When an AI CMO owns the marketing workflow end-to-end — strategy, content creation, publishing, optimization — the CRM becomes a data store it reads from, not the platform the marketer lives in.

The switching cost was never about the data. It was about the human’s muscle memory with the dashboard. Remove the human from the execution loop, and the switching cost evaporates.

Network Effects: Real for Platforms, Irrelevant for Workflows

a16z cites Salesforce’s ecosystem as a surviving network effect: “The more companies use Salesforce, the more valuable the ecosystem of third-party applications built on top of Salesforce becomes.”

True for CRM. Not true for marketing execution.

When you schedule a LinkedIn post through HubSpot, the network effect is zero. The value doesn’t increase because other companies also schedule posts through HubSpot. Unlike Figma — where collaboration creates genuine network value — marketing software is fundamentally a single-player workflow dressed up in a multi-player interface.

The network effects that matter in marketing aren’t in the tool. They’re in the channels: LinkedIn’s network, Google’s search index, your email list. An AI agent accesses those networks directly. It doesn’t need a middleman platform to aggregate them.

Brand: “No One Got Fired for Buying HubSpot” — Until Now

a16z argues that brand endures as a moat, citing the old IBM adage. And they’re right that Stripe and Shopify have trust advantages that AI doesn’t erase.

But marketing software brand works differently. “No one got fired for buying HubSpot” only holds when the alternative is another marketing platform. When the alternative is an AI agent that replaces the labor entirely, the comparison shifts.

You don’t evaluate an AI CMO against HubSpot. You evaluate it against the $200K-$500K marketing team that HubSpot requires to operate. The brand question becomes: “Do I trust this AI to do the job?” not “Do I trust this tool to help my team?”

That’s a different buyer, a different budget, and a different decision framework. HubSpot’s brand advantage doesn’t transfer.

Cornered Resources: Marketing Data Isn’t Bloomberg Data

a16z’s strongest moat example is proprietary data — Bloomberg’s financial data, Abridge’s clinical conversations. These are genuinely cornered: expensive to replicate, legally protected, and compounding in value.

Marketing data looks similar on the surface. HubSpot has millions of campaigns, open rates, conversion benchmarks. But there’s a crucial difference: marketing data is business-specific, not category-specific.

Bloomberg data is valuable because finance is universal — a stock price means the same thing to every trader. Marketing data is the opposite. Your open rates don’t help me. My conversion benchmarks don’t apply to your audience. The aggregate data that HubSpot could theoretically leverage is too generic to drive the specific decisions that matter.

The cornered resource in marketing isn’t aggregate benchmarks — it’s the closed-loop data from your specific brand. Which channels work for you. Which messages resonate with your audience. Which timing converts. An AI agent that owns your workflow builds this cornered resource from day one — and it’s more valuable than anything in HubSpot’s database.

Process Power: The One Moat That Actually Matters

a16z’s most interesting claim: “The hard part was never raw intelligence. It was knowing what to do with it.” Process power — the embedded workflow knowledge that compounds as capabilities expand — is, they argue, the strongest surviving moat.

We agree. And this is exactly where marketing software incumbents are weakest.

HubSpot’s process power is in helping humans use dashboards. Salesforce Marketing Cloud’s process power is in workflow automation rules that humans configure. The entire process is built around human operators.

When the human leaves the loop, that process power doesn’t transfer. It’s like a factory optimized for manual assembly — the process knowledge is real, but it’s the wrong process for robotic automation.

The process power that matters for AI marketing is different: understanding brand voice at a deep level. Knowing when to post and when to stay quiet. Recognizing which competitor move requires a response and which to ignore. Learning a brand’s specific audience over time.

That’s not HubSpot’s process. That’s a Brand Parent’s process.

Counterpositioning: The Moat-Killer

a16z’s analysis of counterpositioning may be the most important section. They cite Decagon, which prices customer support per conversation handled instead of Zendesk’s per-agent-seat model. Zendesk can’t adopt this model without cannibalizing existing revenue — the classic Netflix-vs-Blockbuster dynamic.

The same counterpositioning applies to marketing. HubSpot charges per contact and per seat. Their entire revenue model assumes humans operating the software on a growing database of contacts.

An AI CMO doesn’t have seats. It doesn’t charge per contact. It charges for marketing outcomes — or a flat subscription for autonomous marketing execution.

HubSpot can’t adopt this model. If AI handles the execution, why would customers pay per seat? If the AI owns the workflow, why would customers pay per contact tier? Every step toward autonomous marketing erodes HubSpot’s revenue model.

This isn’t hypothetical. It’s the same structural trap a16z identifies in Zendesk’s position relative to Decagon — applied to the MarTech industry.

The Bifurcation a16z Predicts Is Already Happening

a16z predicts a “great software bifurcation” where vulnerable companies — “thin frontend wrappers around commodity functionality” — lose to companies delivering genuine value.

In marketing, the bifurcation is more specific:

Surviving: CRM platforms (genuine network effects and data gravity), analytics tools (measurement is channel-agnostic), creative tools like Figma and Canva (human taste remains essential)
Vulnerable: Marketing automation platforms, social media schedulers, email campaign tools, content management systems — anything that assumes a human operator executing a marketing workflow

The middle layer — the workflow orchestration that today’s MarTech stack provides — is exactly where AI agents operate. Not above it. Not below it. Right where the incumbents sit.

What This Means for Founders

If you’re building a startup and evaluating your marketing stack:

Your marketing tools are Era 2 software. They store data and help humans work. Era 3 is software that does the work.
Your MarTech budget is paying for human-operated infrastructure. When the human leaves the loop, most of that infrastructure becomes overhead.
The moats protecting your current vendors aren’t protecting you. They’re protecting the vendor’s revenue model from the disruption that would actually save you money.

a16z is right that AI won’t destroy software. But marketing software’s moats are uniquely vulnerable — not because the moats are weak, but because they’re optimized for a world where humans do the marketing. When AI does the marketing instead, the moats protect an empty castle.

Ready to skip the MarTech moat entirely? Get early access to Lane — our AI CMO that owns the marketing workflow, not the dashboard. Or see how it works.

Got questions? Reply to the email that brought you here, or reach out at hello@luminarylane.net.

References

Alex Immerman, Santiago Rodriguez — “Good News: AI Will Eat Application Software” (a16z, Mar 2026)
Hamilton Helmer — 7 Powers: The Foundations of Business Strategy
Related: Software Is Eating Marketing Labor — Our response to a16z’s AI opportunity thesis
Related: Own the Workflow or Lose the Customer — What Harvey and Sierra reveal about workflow ownership
Related: The AI CMO: From Marketing Tools to Brand Parent — Our pillar piece on the Brand Parent category