Claude Code is rapidly taking over the developer market.
According to the latest insights from renowned investor Deedy, Claude Code has attracted an astonishing 115,000 developers in just four months, with an incredible 195 million lines of code rewritten last week!
Based on these figures, Claude Code's annual revenue might already be around $130 million, with each developer contributing over $1000 annually.
Even more surprising is that Claude Code was just launched four months ago, and Sonnet/Opus 4 is only 1.5 months old.
Such growth can only be described as a "meteoric adoption rate."
The True "Junior Engineer"
Deedy stated: "I'm not hyping it up. Claude Code Opus is a junior software engineer."
This assessment is not an exaggeration.
Most AI programming tools on the market are still limited to "smart auto-completion," while Claude Code is already capable of performing tasks typically handled by a junior engineer.
Lekan (@lekan_digital) commented:
195 million lines of code touched in a week, that's insane speed. If Opus is a junior engineer, then it's one who never sleeps, works in 5 shells simultaneously, and already knows your entire codebase. $130M ARR is just the beginning. The real upside is in org-level integration and memory.
Love-Hate Relationship
Interestingly, developers' opinions on Claude Code exhibit a "love-hate" dynamic.
Jake Halloran (@jakehalloran1) joked:
My team's budget wishes I only spent $1000 a year, haha.
This indicates that his spending on Claude Code far exceeds that figure.
Kirk Marple (@KirkMarple) shared a more detailed user experience:
I've used it heavily for the past 6 weeks and unfortunately, those 195M lines of code changes need deduplication — it sometimes performs poorly due to overconfidence, ending up having to fix itself 3-4 times, wasting a lot of tokens.
He added that luckily there's a $200 monthly cap, so costs are acceptable, but wasted time can be terrible — especially when you run out of Opus tokens.
Dan Advantage (@DanAdvantage) empathized:
You're so right! Now I understand the problem! Perfect! I found the problem! Let me test it...
Cursor Faces Pricing Dilemma
While Claude Code is making great strides, its former star product, Cursor, has fallen into a pricing strategy quagmire.
Sankalp (@dejavucoder) revealed:
A major reason for using Cursor, besides the excellent developer experience, was getting $100 worth of API credits for just $20, thanks to VC funding. But I'm not sure if that will continue.
Cursor recently issued an apology statement, admitting their pricing adjustments were "not communicated clearly," and promised a full refund to users who incurred unexpected charges in the past three weeks.
Sankalp showed a comparison between Cursor's original blog post on June 16 and the currently edited version. He said:
I've been working more than usual for the past two weeks, and I used it frantically — it was truly unlimited. I've posted about this twice.
But it didn't last long, as Cursor quickly realized the problem:
Sankalp added:
They must have realized they messed up, haha, they started rate-limiting yesterday.
Cursor's Business Model Predicament
Dushyant (@DevDminGod) provided an in-depth analysis:
Cursor is doomed; in fact, anyone using downstream APIs to compete with the actual model companies will be doomed.
Anthropic is subsidizing their Max plan usage; a $200 account can easily generate several times that usage. Anthropic doesn't care; they can continue operating at a loss because it's not a real loss for them, their input cost per query is just energy and computation.
But Cursor has to pay Anthropic real money. It won't subsidize third-party API usage multiple times over with real money for you.
He also revealed an interesting detail:
Cursor poached the people who built Claude Code from Anthropic, thinking it would be smooth sailing from then on... Haha! No way.
Smart programming consumes insane amounts of tokens — you'll always get a better price from the original company than through a middleman — Cursor's dark days are coming.
Cursor is Losing Ground
Many developers have compared Claude Code with Cursor.
Sivaram (@SivaramPg) directly stated:
If @cursor_ai keeps screwing up, this number will keep rising!
Prashant (@Prashant_1722) even asserted:
Cursor is collapsing.
blek (@yyblek) shared their team's experience:
Our whole team switched to CC + Opus 4 in less than a week. It's currently unrivaled.
Ben (@andersonbcdefg) pointed out succinctly:
It's impossible to compete with Claude Code while using Claude, because Anthropic can sell tokens cheaply. It makes sense that they are now trying to train their own models. If they don't, they will die.
Claude Code's Parallel Capability
Claude Code is not only growing rapidly in user numbers, but its technical capabilities are also impressive.
A Reddit user showed that Claude Code can work on 8 parallel tasks simultaneously, a multi-threading capability that further proves its "junior engineer" qualities.
Even the capability of a junior engineer team.
This parallel processing capability means Claude Code can:
Handle modifications to multiple files concurrently
Switch between different codebases
Execute testing and debugging tasks in parallel
Significantly boost overall development efficiency
Who needs a tech team when you can just open a few Max accounts?
Evolution from Junior to Senior
Developers are generally optimistic about Claude Code's future.
Abhishek Tiwari (@abhishekt282001) predicted:
It is expected to evolve from a junior to a senior-level agent within 6-8 months. Reinforcement learning with scaled developer feedback loops is the closest thing to training a neural system under "pair programming + real-time deployment." This greatly accelerates skill growth.
Noman Khan (@nomcodes) analyzed from an ROI perspective:
If Claude Code can save each developer 5 hours a month, the ROI already exceeds $1000 per year. Multiply that by 100,000+ developers, and you see a serious SaaS giant. And we are still in the early stages.
Francis John (@franjohn21) joked:
What junior engineer makes $130 million/year? That's Soham-level productivity.
Of course, some have raised security concerns.
$ (@fbi__open__up) warned:
What happens when someone hacks these centralized AI code injectors and secretly inserts backdoors into thousands of codebases? It's only a matter of time.
In my opinion, Claude Code is aggressively seizing the digital channel of programming, which is the most likely path to AGI.
Programming is the most versatile tool in the digital world; mastering programming capabilities means mastering the key to creating and changing the digital world.
Anthropic's early bet on AI programming and its current rapid lead demonstrate a deep understanding of the essence of "intelligence."
If anyone can achieve true AGI first, I'm most optimistic about Anthropic.
Because they are not just building a simple code completion tool, but are cultivating an intelligent agent that truly understands code, logic, and creation.
When 115,000 developers rewrite 195 million lines of code weekly through Claude Code, it's clear that Anthropic has a firm grip on the programming domain.
And programming is the most effective bridge connecting the human world and the digital world —
The inevitable path to AGI.