>Their expertise will be used to improve Grok Build coding agent.
Is Grok not a toxic enough brand (by association with Musk) that people who would use Cursor wouldn't avoid Grok?
Like, the assumption seems to be that all the goodwill that Cursor users have towards Cursor will now apply automatically to Grok, which seems like a pretty significant leap.
I've been teaching myself physics lately and have found Grok to be one of the best both at coming to a correct answer and helping me to understand how to get it myself. It also seems a lot better than other models at saying "I don't know" or pointing out when my question doesn't make sense.
I'm just trying to get it to help me learn physics. If my topic of interest shifts to mid-20th century European history I'll keep what you said in mind.
If you're going to use the model to learn history you're going to learn the version of history that the model teaches you. A little bit of digging around grokpedia should give you some idea of what that model thinks
I'm a heavy Cursor user, I spend hundreds of dollars per month in overages above my $200 subscription, and I'll be switching away. I have zero interest in Musk's AI.
The CEOs ideology matters due to the fact it impacts the product design. The reason people don't want to use Grok is because it's bad, and it's bad because the team behind grok have to spend cycles crowbarring in far right white genocide conspiracy theories so that it doesn't embarrass their boss on twitter. One of the things we learned with Anthropic is that you have a lot more success being focused on the core product - coding agent, than trying to do that and chase internet chatbot users.
As someone that trains LLMs, even if grok does have a "avoid ""woke""" fine-tuning, adding that needs a few thousand examples SFT and a system prompt. It costs nothing extra to do it to coding agents and is not the reason why grok sucks at coding. It's just not in the same league in general - it's 0.5T parameters only and not trained specifically for coding at all.
Even if the way they are doing it did damage coding performance, it is a simple matter of serving another model without that fine tuning in the enterprise API preferably only to the grok coding harness (or cursor, now). Coding performance for subscription plans don't move the needle in terms of revenue anyways and quality there doesn't matter as much.
Achieving the improvements from K2.5 -> composer 2.5 with just post-training is actually more impressive. Though I believe their next model is trained from scratch.
> So SpaceX is buying 15 years of future cashflows, assuming no change in revenue. Which is bad assumption given past growth was gigantic.
I'd argue it's a bad assumption in the opposite direction. There's no moat. People can and will switch tooling and Cursor could easily be left with a steep decline in users.
But revenue is not really the informative quantity. If you sell gold you will have a huge revenue, but very little profit. I can be a trillion dollar company too if we exchange dollar bills for face value.
"training good coding models" many would say that is a highly debatable statement, and some would say that is just flat out not true. Cursor has not trained a frontier model from scratch, what they did was take an already made (non-frontier) model and further trained it on their user data about coding outcomes from its coding agent. So, a form of distillation and RL.
Sure - why use cash when you can use bits of paper instead?
I'd expect more of the same to come - good way to lock in some of this crazy SpaceX valuation by converting it into something with a bit more inherent worth.
Cursor is massively overvalued. But so is SpaceX so it all evens out in the end.
> each share of Cursor’s common stock and each share of Cursor’s preferred stock outstanding immediately prior to the Effective Time of the Merger will be automatically converted into the right to receive shares of the Company’s Class A common stock based on an implied equity value of Cursor of $60.0 billion and the price of the Company’s Class A common stock equal to the volume-weighted average closing price thereof over the seven consecutive trading days
Current market cap is 2.66T which is pretty bonkers. Thats about intel, amd, and micron put together.
