To tell whether a technology is growing or dying, count the companies that hire for it each month and divide by the number of companies hiring at all that month. Run that way, the job posts we collected between November 2025 and October 2026 give two clear answers. Cursor, the AI code editor, went from 15.3 to 38.9 companies per 10,000 hiring companies, about 2.5 times as many. Hadoop fell from 21.7 to 13.7 per 10,000, a drop of 37%. Both counts were taken on 10 November 2026.
Which technology is growing and which is dying in 2026?
Cursor is growing and Hadoop is shrinking, and neither answer depends on anyone's opinion of the tools. In October 2026, 6,787 companies that hire for themselves (recruiting agencies left out) posted at least one job that names Cursor, against 1,965 in November 2025. For Hadoop the same count went from 2,790 to 2,396 over those twelve months, while the number of companies posting any job rose by 36%.
That last figure is why the raw count alone misleads. The pool of hiring companies we see grew from 1,284,145 in November 2025 to 1,744,052 in October 2026, so a technology whose count grows by less than 36% is losing ground. Hadoop's raw count fell 14%. Its share of hiring companies fell 37%.
How can you tell growth from hype with data?
Compare each technology's monthly count with the count of all hiring companies, and read the trend across the whole year, not one good quarter.
Cursor's share of hiring companies grew 2.5x in a year while Hadoop's fell 37%
The figure shows companies per 10,000 hiring companies for each technology at the start and end of the year. Cursor rises from 15.3 to 38.9 and Hadoop falls from 21.7 to 13.7.
| Month | Companies hiring for anything | Cursor | Cursor per 10,000 | Hadoop | Hadoop per 10,000 | LangChain | LangChain per 10,000 |
|---|---|---|---|---|---|---|---|
| Nov 2025 | 1,284,145 | 1,965 | 15.3 | 2,790 | 21.7 | 4,733 | 36.9 |
| Feb 2026 | 1,742,308 | 3,489 | 20.0 | 3,246 | 18.6 | 5,393 | 31.0 |
| Jul 2026 | 1,817,088 | 7,107 | 39.1 | 3,070 | 16.9 | 7,506 | 41.3 |
| Oct 2026 | 1,744,052 | 6,787 | 38.9 | 2,396 | 13.7 | 6,223 | 35.7 |
LangChain is the case for reading the whole year. Its raw count climbed 39% from February to July, which is the kind of run that gets called a boom. By October it was back to 35.7 per 10,000, slightly under its November level. On this data, hiring for the framework held steady over the year.
How many companies hire for a framework each month?
The month by month count is the series to watch before you build on a framework. These are the raw counts of companies hiring for Hadoop and LangChain in each calendar month:
| Month | Hadoop | LangChain |
|---|---|---|
| Nov 2025 | 2,790 | 4,733 |
| Dec 2025 | 2,783 | 4,446 |
| Jan 2026 | 3,128 | 5,243 |
| Feb 2026 | 3,246 | 5,393 |
| Mar 2026 | 3,103 | 6,030 |
| Apr 2026 | 2,932 | 6,192 |
| May 2026 | 2,836 | 6,725 |
| Jun 2026 | 2,973 | 7,278 |
| Jul 2026 | 3,070 | 7,506 |
| Aug 2026 | 2,839 | 6,987 |
| Sep 2026 | 2,690 | 6,638 |
| Oct 2026 | 2,396 | 6,223 |
Hadoop peaked in February and has had fewer companies hiring for it every month since July. LangChain peaked in July and has dropped for three months in a row. Each row is one search, and here is October's Cursor count as a message for an AI assistant, an API request and the app's filters:
Using the TheirStack MCP, count the companies hiring for Cursor in October 2026. Look for:
- Jobs that mention the technology Cursor
- Posted between 2026-10-01 and 2026-10-31
- Direct employers only, not recruiting agencies
Tell me how many companies and jobs match, then list the first 10 companies with their job titles.
curl -X POST "https://api.theirstack.com/v1/jobs/search" \
-H "Authorization: Bearer <YOUR_API_KEY>" \
-H "Content-Type: application/json" \
-d '{
"job_technology_slug_or": ["cursor"],
"posted_at_gte": "2026-10-01",
"posted_at_lte": "2026-10-31",
"company_type": "direct_employer",
"include_total_results": true,
"limit": 1
}'- Posted date2026-10-01 to 2026-10-31
- Job Technologyis any of Cursor
- Company TypeDirect employer
With include_total_results and limit set to 1, the API returns the totals for the price of one job, so each month costs 1 API credit (counting results). A year for one technology is 12 credits, plus 12 more for the monthly count of all hiring companies you divide by.
Who hires for Cursor and Hadoop in October 2026?
Hadoop is hired for by bigger companies. Of the 2,396 companies hiring for it in October 2026, 970 (40%) have 1,000 or more employees. For Cursor it is 1,571 of 6,787 (23%), so most of its demand comes from smaller teams.
The United States is about a third of both. 776 of the Hadoop companies (32%) and 2,015 of the Cursor companies (30%) had the job located in the US, so most of the demand for both tools is elsewhere.
Do website technology trackers show back-office software adoption?
Website trackers mostly don't show back-office software, because they read the code a company's site loads in the browser. That shows analytics scripts, e-commerce platforms and front-end frameworks. Hadoop clusters and the editor an engineer writes code in never load on a website, and the same goes for most ERPs, CRMs, databases and data platforms. Job posts name them because the company needs people who can run them.
The two sources cover different ground. BuiltWith, for one, tracks more technologies than we do (more than 124,000 against our 33,000), all found by crawling websites (how the two compare). For a storefront or a JavaScript library, a website tracker is the better source. For back-office software, job posts are where adoption shows up.
How we counted companies hiring for a technology
Each number is a count of distinct companies, run on 10 November 2026 against the job posts we have collected from job boards, applicant tracking systems and company career pages.
- Technology: a job counts when we detected the technology in its title, description or application URL, using our catalog entries for Cursor, Hadoop and LangChain.
- Period: jobs posted in each calendar month, from November 2025 to October 2026. Jobs that have since closed still count for the month they were posted in.
- Companies: direct employers only. Recruiting agencies are left out because they post for other companies.
- Denominator: all direct employers with at least one job posted that month, whatever it names.
The same data sits in the technographics dataset, one row per company and technology with a first_date_found, a last_date_found and job counts for the last 7, 30 and 180 days. Counting rows by the month of first_date_found gives new adopters per month. In October 2026 that was 2,488 companies for Cursor and 429 for Hadoop.
What job post counts can't tell you about adoption
A job post count measures hiring demand, not the installed base, and that leaves gaps.
- Companies that use a tool but aren't hiring for it, or don't name it in their posts, don't appear. A stable Hadoop cluster with no open roles is invisible.
- A post can name a tool the company is moving away from, such as a role to migrate data off Hadoop. A falling technology's count still includes some of its exits.
- Our coverage grew 36% over the year, which is why we divide by all hiring companies. The raw counts in the monthly table overstate growth.
last_date_founddoesn't show who dropped a tool. A company still running it looks like it stopped until it posts again, and companies that rarely post look like churn.first_date_foundis the first time we saw the technology at that company. While our coverage grows, some "new adopters" are companies that are new to our data.- Twelve months is short. LangChain dips in December, and a single year can't separate a seasonal pattern from a trend. Our job history starts in 2021, but volume ramps up over time, so 2023 onward is the safer start for a longer series.
- "Cursor" is also a database term. In the five October matches we read, every one named the AI editor next to Claude Code or GitHub Copilot, but check a sample of your own when a technology's name is a common word.
Run the same count for your technology and market
Open the October search, swap Cursor for the technology you're weighing, and add a country or company size before you change the month.

The screenshot shows the October 2026 Cursor search in the app, with its filters and the company count at the top.
Repeat it for each month and once without the technology, then divide. To run the series from an AI assistant, connect the TheirStack MCP server and ask for one month at a time. Open the Cursor search for October 2026.
Hermes
OpenClaw
Clay
n8n
Make
Zapier
Databar
Salesforce
HubSpot
Attio
Pipedrive
Apollo
Instantly
ContactOut