Anthropic's Super Bowl Ad: Who Won & Lost? | Sierra Hits $150M ARR: Is Customer Support Too Crowded?

The narrative that "software is dead" has dominated recent technology discourse, fueled by rapid advancements in generative AI and fears of shrinking seat counts. However, industry veterans argue this view is reductive and ignores the historical resilience of the sector. In a recent discussion featuring Atlassian Co-Founder Mike Cannon-Brookes alongside investors Jason Lemkin, Rory O'Driscoll, and Harry Stebbings, the panel dissected the current state of SaaS. From Anthropic’s massive revenue projections to the "consensus winner" phenomenon driving valuations like Harvey’s, the conversation offers a pragmatic look at how enterprises are actually budgeting for and implementing AI.

Key Takeaways

• SaaS is evolving, not dying: While some categories face seat compression, engineering and product sectors remain "islands of stability" where AI accelerates creation rather than replacing roles.
• The "Consensus Winner" premium: Venture capital is flocking to perceived market leaders (like Harvey and Anthropic) regardless of valuation, driven by a psychology that favors safety in numbers.
• Customer support is bifurcating: The industry is splitting between legacy incumbents and new AI-native agents, with internal helpdesks offering massive efficiency opportunities.
• Public vs. Private disparities: Public companies face strict scrutiny on profitability, while private AI startups burn capital on brand plays like Super Bowl ads without immediate ROI pressure.
• Founder resilience is non-negotiable: In a market defined by rapid disruption, founders must find genuine enjoyment in the "game" of business to survive the increased workload.

The "Death of Software" Narrative vs. Reality

The prevailing fear in the market is that AI will decimate traditional software revenue models by reducing the number of human seats required. However, the panel suggests that this view conflates different types of software and ignores the necessity of pre-packaged solutions.

• The "Above the Fold" distinction: The panel introduces a critical distinction between software that serves "builders" (engineering/product) and software that serves "operators" (support/sales).
• Engineering remains stable: Mike Cannon-Brookes notes that companies are not shrinking their engineering teams; instead, they are using AI to build significantly more software.
• Seat compression is localized: Categories like customer support and rote administrative tasks face "existential risk" of shrinking seats, while product management and coding tools see increased demand.
• Complexity requires management: As AI generates more code and products, the need for tools to track, manage, and orchestrate that complexity (like Jira or Confluence) actually increases.
• The efficiency argument: Businesses will continue to buy SaaS because it is inefficient to build custom internal tools for every process, even with AI assistance.
• Historical context: The industry has survived multiple platform shifts; the current "death" narrative is likely just a sectoral rotation where new winners emerge while old ones fade.

The idea that software as a category is dead is ludicrous to me.

Anthropic, Revenue Stacking, and the Hunt for Budget

With Anthropic projecting massive future revenue, the conversation turned to where this money will come from. The debate centers on whether AI spend is additive or if it must cannibalize existing IT budgets.

• The zero-sum game debate: If AI companies hit their multi-hundred-billion-dollar targets, the revenue must either come from TAM (Total Addressable Market) expansion or be taken from incumbents like Microsoft.
• Revenue stacking dynamics: Cannon-Brookes explains that spending on Anthropic often flows through cloud providers (AWS), creating a complex "margin stack" rather than a direct revenue line.
• Consulting as the "honey pot": The panel identifies the trillion-dollar consulting and services market as a prime target for software to eat, potentially freeing up budget for AI tools.
• Implementation costs: Paradoxically, the initial phase of AI adoption drives more consulting revenue (e.g., Accenture) as enterprises need help integrating these new systems.
• Corporate budget elasticity: Technology budgets have historically expanded over the last 30 years; AI is expected to continue this trend rather than forcing a hard cap on spending.
• Inference costs are dropping: Contrary to fears of spiraling costs, Atlassian reports that inference costs are decreasing due to infrastructure optimization and model selection.

The Logic Behind Harvey’s $11 Billion Valuation

Legal AI platform Harvey recently raised funds at a staggering valuation, sparking debate about "wrapper" startups and venture capital psychology. The consensus is that traditional valuation metrics have been temporarily suspended for perceived category winners.

• The "Consensus Winner" effect: Investors are piling into companies like Harvey because they are viewed as the "IBM" of their category—safe bets that justify high entry prices.
• Ignoring traditional multiples: With Harvey valued at roughly 50x forward revenue, investors are betting entirely on massive compounding growth rather than current financials.
• The irrelevance of "wrappers": The criticism that a company is just a "GPT wrapper" is dismissed; if the product solves a high-value problem and has sticky customers, the underlying tech stack matters less.
• TAM expansion in law: The bullish case for Harvey relies on the software doing more than just aiding lawyers—it must eventually perform the billable work itself to justify the valuation.
• Input constraints: Legal work is viewed as "input constrained" (there are only so many lawsuits/contracts), meaning efficiency gains might cap total revenue unless the software opens new markets.
• Scarcity value: Part of the valuation premium comes from the fact that there are very few high-growth, high-quality AI assets available for late-stage private investment.

It's the venture capital equivalent of you can't get fired for buying IBM.

Is the Customer Support Market Too Crowded?

With startups like Sierra reaching $150M ARR and incumbents like Salesforce and Atlassian launching their own agents, customer support is the most contested battleground in AI. The panel dissects whether this market has room for everyone.

• Bifurcation of the market: The sector is splitting into external support (B2C interactions) and internal helpdesk (B2B/employee workflows), each requiring different technical approaches.
• The hiring freeze: Data suggests that customer support roles have seen the most significant hiring slowdowns, confirming that companies are swapping headcount for software.
• Jevons Paradox in support: Cannon-Brookes argues that as support becomes cheaper and better via AI, the demand for support interactions will likely skyrocket rather than plateau.
• Action-oriented agents: The next value unlock isn't just answering questions (text generation) but taking actions (resetting passwords, processing refunds), which favors vendors with deep integrations.
• Internal knowledge gaps: AI agents are forcing companies to improve their internal documentation, as agents require structured data to function effectively.
• Incumbent advantage vs. agility: While startups move fast, incumbents can roll out "good enough" AI features to millions of existing customers, potentially stifling standalone competitors.

Public Discipline vs. Private Excess

A recurring theme in the discussion is the diverging realities of public SaaS companies and private AI startups. This bifurcation affects everything from marketing spend to strategic planning.

• Super Bowl signaling: The panel views Anthropic’s Super Bowl ad not as a mass-market play, but as a high-priced signal to a small group of engineers and investors in Silicon Valley.
• Capital efficiency: Public companies are penalized for declining margins, while private startups are currently rewarded for growth at all costs, leading to distorted competitive behaviors.
• The "Ego" factor: High-profile ad spends often stem from founders with excess capital and ego, rather than calculated ROI decisions—a luxury public CEOs rarely have.
• Strategic R&D: Public companies must balance delivering quarterly earnings with massive AI R&D investments, a "double duty" that private competitors don't face to the same degree.
• Investor confusion: Public market investors are struggling to identify who the AI winners will be, leading to volatility and "fear-driven" pricing for traditional SaaS stocks.
• The inevitable normalization: History suggests the current period of "free money" behavior in private AI will eventually correct, bringing discipline back to the sector.

Founder Resilience and the Mental Toll

The conversation concluded with a candid look at the human cost of the current technology shift. For founders and CEOs, the AI era demands a new level of intensity and adaptability.

• Acceptance of reality: Successful leaders are those who stop mourning the "easy" growth of 2021 and accept that the current environment requires harder work and new strategies.
• The necessity of enjoyment: Given the increased workload and stress, founders who do not genuinely enjoy the act of building and problem-solving are advised to step aside.
• Balance as a performance enhancer: Cannon-Brookes argues that spending time away from work (e.g., with family) isn't a distraction but a necessity for making high-quality decisions.
• The difficulty of quitting: It is often harder for a founder to admit they are burnt out and leave than to stay and perform sub-optimally; knowing when to fold is a skill.
• Creation involves destruction: Leaders must be comfortable with the fact that adopting new technologies may require destroying old business models or processes.
• The long game: With the market in flux, maintaining a long-term perspective (5+ years) is the only way to navigate short-term volatility and narrative shifts.

Part of creation is destruction. You got to go build value for customers.

Conclusion

The "death of software" is less a funeral and more of a metamorphosis. As Mike Cannon-Brookes and the panel highlighted, the industry is undergoing a violent rotation where capital, labor, and value are shifting from rote tasks to creative and complex problem-solving. While the transition creates uncertainty—manifesting in confused stock prices and defensive Super Bowl ads—the fundamental demand for technology to solve business problems remains the world's most powerful secular trend.