The Rundown — October 26, 2025

Episode: The Slow Burn of Tariffs & AI's Infiltration of Corporate America (ft. Ara Kharazian)

Hosted by Zaid Admani | Guest: Ara Kharazian (Economist, Ramp)

Overview

In this episode, Zaid sits down with Ara Kharazian, an economist at Ramp, to unpack emerging business trends drawn from the company’s massive real-time expense dataset. Topics range from the real adoption of AI across sectors (is it all hype, or are U.S. businesses truly committing dollars?), to the complicated rollout and impacts of tariffs, and the ongoing debate about government data’s role versus private sector insights. The conversation illustrates how actual business spending paints a nuanced story about both headline trends and the quirky realities of “everyday” corporate America.

Key Discussion Points & Insights

1. About Ramp & Ara Kharazian's Role

• Ramp’s Dataset: 45,000+ diverse U.S. businesses’ expense data (outside payroll).
• Why It Matters: Business spend is often a leading indicator—offering unique economic foresight not captured in traditional consumer-focused stats.

  “Business spend can often be a leading indicator of where the economy is headed and that data is just not available anywhere else.” (B, 01:38)

2. AI Adoption: Bubble or Real Transformation?

Data-Driven Insights vs. Hype (02:41–07:09)

• Retention is Up:
  • 80% of companies keep AI subscriptions after a year now, up from 50% in 2022 (pre-ChatGPT).

    “If you are a company, you’re buying an AI product or service, you are seeing benefits of it throughout the year such that you are extending your contract…” (B, 04:17)

• Contract Size is Exploding:
  • 2023: The typical AI contract was $30–$40k/year.
  • 2025: Now $500k/year, with $1M/year expected in 2026.

    “That tells us that the investments in AI are not just seen as small pilots. These are large scale investments by organizations…” (B, 05:32)

• Meaningful Productivity:
  • AI is being integrated into customer service (e.g., automating many support tasks) and software engineering, delivering cost savings and freeing up human focus.

3. Who’s Winning the AI Race? (07:09–11:07)

• OpenAI Dominates:
  • Adopted by 30–40% of Ramp’s businesses.
  • Anthropic (10–20%): Competitive with tech companies, especially for large API spend.

  “OpenAI is by far at more companies than any other AI model company.” (B, 07:36)

• More than Model Providers:
  • Some of the biggest spends aren’t on OpenAI/Anthropic/Gemini but on enterprise tools:
    • AI customer service agents
    • Trust/observability platforms
    • Coding tools tailored for enterprises

    “...the largest contracts we often see are for things like companies building a dedicated AI customer agent software.” (B, 06:20)
    “A lot of the platform investment…is happening outside of the sort of big three.” (B, 11:11)

4. Sectoral AI Adoption: Unexpected Leaders (12:16–13:44)

• Leaders: Tech, finance.
• Fastest Growth: Healthcare and construction/manufacturing.
  • Use: Planning, automation, design (not massive robotics yet).
  • In healthcare, AI is automating administrative tasks (e.g. doctors’ note-taking).

  “The fastest growth is coming from two sectors I think are actually unexpected: healthcare and then construction, manufacturing.” (B, 12:28)

5. Is AI Displacing Jobs or Boosting Productivity? (14:03–16:39)

• Early Research: Too soon to quantify workforce impacts.
• Anecdotes:
  • Tech CEOs observe revenue growth “decoupling” from headcount expansion.

  “…there has been this decoupling from company growth to headcount growth…able to grow their revenues without dramatically increasing headcount.” (B, 14:54)

• Big Picture: AI likely to spur new kinds of work, as other past technologies have.

  “We seem to keep finding other stuff to do. And I'm not a doomer on this stuff.” (B, 15:53)

6. Tariffs: Expectations vs. Reality (17:17–23:03)

• Rollout is Slow & Patchy:

  • Expected big inflation didn’t materialize.
  • Only 3% of bills/invoices show a tariff charge (as of September 2025, up from 1.4%).
  • Many tariffs delayed or renegotiated; actual enforcement is complex.

  “It's not like, oh, the tariffs were announced and then immediately they started getting collected at the ports.” (B, 19:19)

  • Businesses shifting suppliers or waiting for clarity.

    “…many businesses are changing their suppliers…Many of them are moving supply chains to countries that are relatively low tariff.” (B, 21:12)

• Future Outlook:

  • Tariffs may reach low double digits (10–11%) at most, but trends are hard to forecast given lags and policy changes.

  “But it is really hard to say given how laggy the policy is.” (B, 22:42)

7. Government Data vs. Private Data: What’s Reliable? (23:03–28:56)

• Criticism of Gov’t Data: High-profile revisions, methodology debates, private sector data seen as “real-time.”
• Why Gov’t Data Still Matters:
  • Best for covering tough-to-reach Americans, especially those outside big-company payrolls.
  • Critical for benchmarking and triangulating private data.

  “Private sector data sets will never replace government data sets.” (B, 24:06)

• Complementary Roles:
  • Private sector offers new lenses (e.g. AI spend via Ramp).
  • Gov’t excels in macro data (inflation, unemployment).

  “…my main goal is: can I produce work that is helpful and additive…that does something that the government…cannot adequately measure…” (B, 27:39)

8. Quirky Corporate Spending & Cultural Data Nuggets (29:57–33:48)

• Weirdest Corporate Expenses:
  • Halloween costumes, concert tickets, apologetic explanations for big Uber bills.

  “The funnier thing I see is that when people use Ramp to do their expenses…sometimes people are unusually apologetic…” (B, 30:13)

• Restaurant Rankings:
  • The Smith (NYC) tops corporate dining, sparking debates online.
  • Restaurant posts are Ara’s most popular analytics—"nobody cares about economics, they just want to know what restaurant to go to." (B, 32:45)
• Who parties the hardest?
  • New York leads in corporate alcohol and party spend, with New Orleans also a standout.

  “Quantifiably New York…one of the only cities where people are drinking more in the last couple of years. And that share just continues to grow up.” (B, 33:12–33:48)

Notable Quotes & Moments

• On AI contract size and stickiness:

  “We're now seeing average product size of about $500,000 a year. And we estimate that the average AI contract will hit $1 million next year.” (B, 05:19)

• On sector surprises for AI adoption:

  “Healthcare and then construction, manufacturing…are adopting AI faster now than we expected.” (B, 12:27)

• On government data necessity:

  “Private sector data sets will never replace government data sets…They are both important as a benchmarking tool and in helping me triangulate my results.” (B, 24:06)

• On quirky expense memos:

  “The funnier thing I see is that…someone is extremely apologetic. ‘I really looked through all of my available options and this is the only thing I can do…’” (B, 30:13)

Timestamps for Key Segments

• [01:05] Ara introduces Ramp, describes his economist/data analyst role
• [02:41] Is AI a bubble or not? - Data on AI spend and retention
• [05:32] AI contracts are rapidly increasing in dollar size
• [07:36] OpenAI’s dominance in business AI spend
• [12:27] Healthcare and construction lead in AI adoption growth
• [14:54] Productivity gains vs. job loss fears in AI adoption
• [17:17] Why tariffs’ effects have been milder and more gradual than expected
• [23:44] Trust (or lack thereof) in government data, and its role relative to private data
• [29:57] Weird corporate expenses and consumer behavior trends
• [33:12] Which cities spend the most on parties and drinking

Tone & Style

The conversation is lively, data-driven, and peppered with real-world anecdotes that humanize big trends. Ara’s tone is measured, optimistic about technology, and advocacy-based when it comes to the continued value of government data. Zaid keeps the conversation accessible and injects humor, especially around the oddities of corporate spending behavior.

Takeaways

• AI adoption in corporate America is both real and accelerating, not just hype: retention rates and contract sizes are rising dramatically.
• “Bubble” concerns about AI are overblown—actual business spend and practical application suggest sustained value.
• Tariffs are having a slower, more localized impact than headlines suggest; administrative/law enforcement lags are blunting their immediate effects.
• Government data remains vital for full economic understanding, despite private datasets’ speed and specificity.
• Business expense data uncovers unexpected social trends—from restaurant popularity to city-level party cultures.

Guest plug:
Ara’s data-driven writing: econlab.substack.com
More: Ramp Economics Lab | Twitter: @arakharazian

Summary by The Rundown for investors and the economically curious — October 26, 2025