The biggest gap between startups and enterprise we've ever had

I think the gap between startups working with Opus 5.5 level models and enterprises is the biggest gap between startups and enterprise that we have ever had.

Why big companies move slowly

Enterprises have always had to be more risk averse. They have clients to lose, compliance and security blockers, certifications (I wrote about how few agent tools get through a due-diligence questionnaire in October), and that's a lot of friction aimed at keeping things stable and not changing that much. In the end you're a big company, you have product market fit and you don't want to destroy that. Clayton Christensen wrote the book on this in 1997, and in The Innovator's Dilemma the well-run incumbents lose by listening to their best customers, who don't want the new thing yet, while the new thing grows up in a market too small to move a big company's numbers ^[1].

On top of that add the bigger org, the processes around it and more people to update. Fred Brooks put a number on that in 1975, and n people have n(n-1)/2 lines of communication between them, so a team of 10 has 45 and a team of 50 has 1,225 ^[2].

And on top of that there's shareholder value and the enshittification of the product, which is why they lost money to gain market share, so they could squeeze later. Cory Doctorow gave it a name and set it out in January 2023. A platform is good to its users first, then squeezes its users for its business customers, then squeezes the business customers to take everything for itself ^[3]. All of this is known and has been discussed for years.

What changed

What I've noticed, working on both sides, is that the gap between enterprise and startups has been getting bigger and bigger. The whole "SaaS is dead" trend (it isn't, you still need sales and marketing and clients) did have some truth in it. Satya Nadella gave it a big push in December 2024, when he said on the BG2 podcast that business applications would probably collapse in the agent era, as CX Today reported ^[4]. In February 2026 software stocks went through what Fortune's headline called a trillion-dollar selloff, after Anthropic released industry plugins for Cowork ^[5]. The truth in it is that you can build it for a fraction of the cost compared to last year, and a fraction of a fraction compared to three years ago.

What has changed even more in the last year is that you can build it much better, and at that fraction of the cost. Anchor Terminal got to two million page views in about a month with two of us and two Claude subscriptions, and it's past three million since launch now, at about 200,000 a day ^[6]. My previous business took two years and a team of ten to get to two million.

This won't be seen for a few more years, while licences run out and companies die, but it will be a big shift to smaller tech companies, companies with fewer than 50 engineers that are winning their space. Look at Oracle. It's still alive somehow, even though it's had an inferior product for over 25 years, and it made $67.4 billion in its 2026 financial year, $24.5 billion of that still from software ^[7]. So it will take time, but it will eventually hit.

Start now

So my advice is to find your niche, do that thing you're passionate about and see it blossom and grow. Do it while you still have the engineering job, before you hit the jobs market at the same time as 200k of your peers and everyone is scrambling for work. (The worst year for tech layoffs so far was 2023, with 265,660 by Second Talent's count of the layoffs.fyi data, and 2026 had passed 128,000 by 10 September ^[8].) Do it now. It costs a Claude subscription, or whatever model and harness you choose, but it will be your future job.

References

Read on 9 November 2026.

1. [1]Wikipedia, The Innovator's Dilemma (Clayton M. Christensen, Harvard Business School Press, 1997). Source for the book's argument that well-managed incumbents lose to entrants by serving their best customers, and that small new markets can't meet a big company's growth needs. en.wikipedia.org
2. [2]Wikipedia, The Mythical Man-Month (Fred Brooks, Addison-Wesley, 1975). Source for Brooks's law and the n(n-1)/2 formula for lines of communication in a group, with 50 developers giving 1,225. en.wikipedia.org
3. [3]Cory Doctorow, Pluralistic, Tiktok's enshittification, 21 January 2023. Source for the three stages of platform decay. pluralistic.net
4. [4]CX Today, Microsoft CEO, AI Agents Will Transform SaaS as We Know It. Source for what CX Today reported Satya Nadella said on the BG2 podcast with Bill Gurley and Brad Gerstner in December 2024, that business applications could collapse in the agent era and that they are CRUD databases with business logic. cxtoday.com
5. [5]Fortune, Beatrice Nolan, Anthropic's Claude triggered a trillion-dollar selloff. A new upgrade could make things worse, 6 February 2026. Source for the selloff in enterprise software stocks after the industry plugins for Claude Cowork, and for the headline's trillion-dollar figure. fortune.com
6. [6]Anchor Terminal's own request records, page views of every kind (people, agents and crawlers) since the site started, and the daily rate, as of 9 November 2026. Not published in full; what the records keep is on the privacy page. anchorterminal.com
7. [7]Oracle, Oracle Announces Record Q4 and FY 2026 Results Driven by Cloud Infrastructure & Cloud Applications, 10 June 2026. Source for fiscal 2026 total revenues of $67.4 billion, cloud revenues of $34.0 billion and software revenues of $24.5 billion. oracle.com
8. [8]Second Talent, tech layoffs statistics, September 2026, a tally of the layoffs.fyi database. Source for 265,660 tech employees laid off in 2023 and 128,536 in 2026 to 10 September. secondtalent.com