DOP 374: Does AI Make Companies More Productive?

October 28, 2026 Episode 374

Show Notes

#374: Your company is paying for the AI licenses, and now there’s a number attached. Two to three times the output, every day, or else. Can you hit it? If they’re counting lines of code, sure.

But you aren’t the company. If everyone on the team gets faster on their own and the handoffs between them stay exactly the same, does anything ship sooner? Most AI today is single player. Your company probably isn’t.

In this episode, we speak with Tomasz Tunguz, founder of Theory Ventures, about why faster individuals don’t add up to a faster company, and what that changes for startups and the people funding them.

Frequently Asked Questions

Does AI make companies more productive?

Not yet at the company level, according to Tomasz Tunguz of Theory Ventures on DevOps Paradox episode 374. He argues most AI today is single player: it clearly raises individual productivity, but those gains don’t necessarily add up to enterprise productivity, because each person is optimized in isolation rather than the whole system. Viktor Farcic agrees individuals are much more productive, but doubts the gain survives the handoffs inside a team or company.

How much more productive does AI make people?

Tomasz Tunguz, founder at Theory Ventures, says on DevOps Paradox episode 374 that the studies he has seen put the overall gain at 25 to 150 percent on the high end, but closer to 20 to 50 percent today. His own blog shows a different kind of gain. He still makes roughly 130 edits per AI-assisted post, but when he graded fifteen posts from each of the last ten years, this year’s quality had surged even though he was not writing more posts.

Are AI startups hiring junior engineers?

Most new startups begin with senior engineers and an architect who design the software factory, according to Tomasz Tunguz on DevOps Paradox episode 374. They typically don’t hire juniors until the team reaches about 15 to 20 people. He says training juniors is unsolved, because assigning them PRs to review fails when they lack the experience to judge them. His firm, Theory Ventures, has seven AI engineers and hires technical interns while it works out how to teach them.

Is innovation invention plus distribution or invention times distribution?

Darin Pope argues on DevOps Paradox episode 374 that innovation should be invention times distribution, because with zero distribution you have nothing. Guest Tomasz Tunguz had quoted an equation from a Belgian marketing professor, innovation equals invention plus distribution. He accepted Darin’s refinement and said he would update a blog post to make the relationship multiplicative. Tomasz adds that as the cost of inventing falls, more of the weight shifts toward distribution.

Can the US build enough data centers for AI by 2030?

Tomasz Tunguz says on DevOps Paradox episode 374 that the US won’t get there, because supply chains, electrical grids and politics are not ready. He puts current US data center capacity at roughly 20 gigawatts, mostly in Loudoun County, Virginia, against projections of 200 to 300 gigawatts by the end of 2030. Concrete, the most expensive component, now costs over three times what it did two years ago. Darin Pope calls the projection impractical.

What is the token path for AI startups?

As Tomasz Tunguz describes it on DevOps Paradox episode 374, the token path is the position a startup needs to grow at today’s rates. That means reselling inference in some form, packaging inference and AI into its product, or providing the tools that let inference be sold. His example of the last is Twilio, which sells no voice inference but connects a voice agent to a person by phone. He says this is why the per-seat software model is dying.

What is the DevOps Paradox podcast?

DevOps Paradox is a weekly podcast co-hosted by Darin Pope and Viktor Farcic, covering DevOps, platform engineering, and modern software delivery. Episode 374, with Tomasz Tunguz of Theory Ventures, asks whether AI makes companies more productive, why individual gains don’t yet add up for enterprises, and what the AI boom means for startup funding and data centers. Every episode page carries audio, video, and a full transcript.

Episode Transcript

Read the Full Transcript

Guests

Tomasz Tunguz

Tomasz Tunguz is the founder of Theory Ventures, investing in early-stage software companies. Previously, Tomasz was a Redpoint Ventures GP and a product manager at Google, managing a billion dollar business unit on the AdSense team. As an investor, he’s worked with eight unicorns predominantly in data and data infrastructure, including Looker, Monte Carlo, Dremio, Hex, Omni, and Mother Duck. He writes a blog at tomtunguz.com that drives millions of page visits per month — it features insights gleaned from benchmarking, interviewing, and analyzing private and public market companies.

Hosts

Darin Pope

Viktor Farcic

Viktor Farcic is a member of the Google Developer Experts and Docker Captains groups, and published author.

His big passions are DevOps, Containers, Kubernetes, Microservices, Continuous Integration, Delivery and Deployment (CI/CD) and Test-Driven Development (TDD).

He often speaks at community gatherings and conferences.

He has published DevOps Paradox and Test-Driven Java Development.

His random thoughts and tutorials can be found in his blog The DevOps Toolkit.