Skip to Main Content

JDOT #7: Cursor vs. Claude Code

Joe Devon

July 6, 2025

Cursor and Anthropic's battle over AI coding, Joe's experience with AI during an illness, and news from the Accessibility and Gen AI Podcast.

Two muscular arms with AI coding logos arm-wrestle above the word Claude; a doctor and medical scans, a podcast microphone, and an accessibility symbol appear in the background.
Alt: 2 muscular arms with tech logos arm-wrestle over "Claude"; a doctor with scans; a podcast mic with AI & accessibility themes. Image by Gemini.
View full-size cover image

The AI Wars take a page out of the Streaming Wars

The pivotal moment in the Streaming Wars came in January 2013, when Netflix Chief Content Officer Ted Sarandos said "The goal is to become HBO faster than HBO can become us."

Ted Sarandos: The goal is [for Netflix] to become HBO faster than HBO can become us.

In 2025, Netflix makes far more content than HBO does. And where is HBO? It has rebranded so many times that I don't know if they are HBO again, or MAX or HBO Max.

History is repeating itself with the AI Wars. While many are distracted by Mark Zuckerberg poaching what's left of the core OpenAI researchers with offers reportedly in the $100M-$300M range, don't sleep on what is currently an existential AI battle between the fiercest of frenemies: Cursor vs Anthropic.

Two spy-like cartoon figures face off in front of the Golden Gate Bridge. The black-clad figure labeled "Cursor" holds a laptop reading "Biggest Coding Dataset." The red-clad figure labeled "Anthropic" holds a tablet reading "Claude World Champion AI Coder." Text above reads "Cursor must become Claude..." and below, "...before Claude becomes Cursor." View full-size image: Two spy-like cartoon figures face off in front of the Golden Gate Bridge. The black-clad figure labeled "Cursor" holds a laptop reading "Biggest Coding Dataset." The red-clad figure labeled "Anthropic" holds a tablet reading "Claude World Champion AI Coder." Text above reads "Cursor must become Claude..." and below, "...before Claude becomes Cursor."

AI-powered coding represents LLMs' biggest success story. The market leader is Cursor, an AI coding assistant and the fastest growing SaaS product ever.

Their "best friend" is Anthropic, the company behind the Claude family of large language models (LLMs), because Claude models are currently SOTA ("State-of-the-Art") at coding tasks.

Anthropic's best friend is Cursor, because Cursor showcases Claude's coding superiority to the world. This matters deeply to Anthropic who lost the mindshare battle to ChatGPT by prioritizing safety over speed to market when launching their consumer chat product.

It isn't just branding though. Anthropic earns significant revenue from API calls that Cursor makes to their models.

So why would Anthropic release Claude Code, their own AI coding assistant, competitive to their best customer?

But wait! Cursor just poached Boris Cherny, who created the Claude Code product! What the heck?

To make matters more interesting, over July 4th weekend, Cursor users have been apoplectic on social media. They (and I) have been either rate-limited or hit with enormous bills after switching to Cursor's new pricing plan, which was supposed to be unlimited but turned out to be worse than before. On Reddit, there's a trending thread called "The great unsubscription."

Someone on social media plausibly speculated that Anthropic started throttling Cursor in anger over the employee poaching. You would think Cursor's deals with Anthropic are backed by solid contracts, but you never know what's really happening behind the scenes.

What is at the heart of this battle? What fundamental issue pits these two frenemies against each other?

Simple: AI models thirst for data. The best and most voluminous data creates the best models.

Cursor has a "tab" feature that adds or edits AI code inline as you type, like autocomplete on steroids. When you accept a suggestion, that's valuable training data. When you prompt the chat to create a new feature and accept the resulting code, that's literal gold for model training.

With the possible exception of Microsoft-owned GitHub with their GitHub CoPilot product, Cursor is best positioned to mine the data needed to create the world's best AI coding model. But until they do, they remain dependent on Anthropic.

This mirrors how Netflix was in the precarious position of depending on the very studios they were competing with for content. Netflix also had to worry that the networks, seeing Netflix's success, could enter Netflix's business and squeeze them out of needed content. Netflix HAD to make original content to protect their future.

The same dynamic applies here. That's why Anthropic doesn't want a layer between them and their users. And why Cursor better be working on creating their own models before their users migrate to Claude Code.

Ironically, I finally tried Claude Code again for the first time since the beta, and the product really does live up to the user buzz. I love it.

Who will win this war? My bet is on Cursor if they become Claude before Claude becomes Cursor, and Claude if the reverse happens first.

I'll finish this section with my current go-to stack:

I use Claude Code at the $100/mo Max level, running inside Cursor on a $20/mo plan. I also use it on a $15/mo plan inside WindSurf, another AI coding assistant and Cursor competitor that's reportedly being acquired by OpenAI.

AI and Your Health

On a personal note, you may have noticed that I haven't answered your DM or email or been around as much. That's because I spent May and June dealing with poor health. I've been managing multiple issues stemming from one foundational problem. I dropped a lot of balls and lost a potential new client.

Fortunately, after the foundational problem returned with a vengeance, my face bloated to the point that I looked like a UFC fighter who just got beat up by Khabib, my doctor finally figured out what was wrong and put me on the right medication.

AI has been incredibly helpful through this process. One of my closest friends is a doctor, and he told me years ago that the "system" has gotten so bad that if he spends more than 10 minutes with a patient, he loses money.

With AI, you can calmly document your medical history, list every symptom, share ingredient lists from food you buy, detail your menu and medications, and it never gets tired of you.

AI was the first to suggest what turned out to be the correct diagnosis: severe idiopathic angioedema. It saved me from severe dehydration. That said, it did NOT replace my doctor, and I'm thrilled to have both AI assistance and professional medical care. And to be in better health.

I'm back to building again, and can't wait to share it all with you, right here.

Accessibility and Gen. AI

Since the last JDOT, we've released a fun new episode of the Accessibility and Gen AI Podcast, Episode #9, with Matt Ater, SVP at Vispero®.

Watch on YouTube: Episode 9 with Matt Ater, Senior Vice President at Vispero

We talked about the Kiosk Hall of Fame, some of the items in Vispero's roadmap and dumb things sighted folks say to blind people. Let us know what you think!

In a couple of days, we will release Episode #10 with Josh Miller, Co-CEO of 3Play Media. And next week we are recording Episode #11 with my friend Victor Tsaran.

Until next time!