Ex-Credit Suisse, Cisco corp dev, VC, startup CFO. Mostly tech and markets, sometimes running and games.
Joined January 2013
- Tweets2.2K
- Following853
- Followers696
- Likes18.7K
Wow Kotaku is calling it "stolen labor" because the developer of Big Walk used AI “to help work through difficult technical challenges and bugs.”
Get some friends together and go play the game, it's awesome. I fully endorse their use of AI, heck, I want them to use more AI to add more puzzles and catch more bugs.
House House confirms 'under 0.1 percent of Big Walk's code' was written using AI following IGF controversy
kotaku.com/house-igf-awards-…
1
57
Feels like I see an awesome post like this everyday about someone vibe-coding a passion project with friends on Opus + Tripo/Meshy + Blender.
I imagine it’s scratching a similar creative itch and desire for fun and connection that people who build StarCraft custom games or Roblox world have.
One of the commenter in the post cautioned to OP that the game needs a strong personality.
I think we’re going to get flooded with a lot of similar looking games soon. And then lots of crazy new things with “personality” that might work. They get copied. Market get saturated. And repeat. Unless the market grows, there’s a diminishing opportunity every cycle.
As a user, discovery is already a problem. And it’s going to get so much worse.
I will probably just stick with Nintendo and play Zelda anyway.
/rant
1
2
136
$RBLX Roblox CEO: "There is a massive surge in creation velocity happening right now on Roblox. And I'll share a few stats. This month, daily asset uploads to our Roblox Cloud have tripled month-to-month. And we think assets, with our images, textures, meshes, are a good proxy for the amount of creativity that's going on. Our game volume leveraging AI has grown 50% quarter-by-quarter. We're seeing some of these experiences now become AI-driven hits."
1
15
How come Claude doesn't do this yet? They must think their compute is better used elsewhere still.
Building foundation models for image/video generation is like going after an entirely new vertical.
The opportunity that exist in front of them with coding, agents, and core reasoning is too big to divert more resource towards image/video.
OpenAI killed its viral video app in March because it was burning through a reported $1 million a day.
Today they snuck the best parts of it back into ChatGPT, for everyone, using a file format from 1987.
The app was Sora. You might remember it from last fall, when AI videos of real people doing ridiculous stuff were all over your feed. That was the cameo feature. You recorded your face once, and Sora could put you in any video. It hit #1 on the App Store and got to a million downloads faster than ChatGPT did.
By January it had fallen out of the top 100. Disney had signed on for a $1 billion partnership, and that died with it.
Now look at what OpenAI shipped today. You can type a prompt or upload a photo and get back an animated GIF. You can also save pictures of yourself, your dog, or a character once and pull them into anything you make later.
If that second one sounds familiar, it should. That's the cameo feature, just for pictures this time.
The GIF is where OpenAI got clever. CompuServe came up with the format in 1987 so people could send pictures over dial-up. Each frame maxes out at 256 colors, and there's no way to attach sound. OpenAI's developer docs cap these GIFs at 64 frames.
A Sora 2 clip could run 10 seconds of full-color video with sound synced to the picture. That's a few hundred frames that all have to look like the same scene, plus a whole soundtrack to generate on top. Every limit baked into the GIF format strips out something OpenAI used to pay for.
And this time nobody has to download a new app. ChatGPT already has about a billion people using it every week. OpenAI says its image tools now make 4 billion images a week, up from 3 billion just two months ago.
The Sora API shut off for good on September 24. Forty-one days later, animation was back as a button in the chat box.
The best ideas from a $1-million-a-day app came back as a GIF maker with the sound turned off.
1
47
Private tech valuations in the spotlight after public market correction axios.com/2026/11/06/private…
Stephen Chou retweeted
Introducing Personal Finance in Claude, in beta in the US.
Connect the accounts you choose, and Claude answers your money questions in everyday language, from your own balances and transactions. Access is read-only, and Claude can't move, manage, or invest your money.
Stephen Chou retweeted
Writing this in my personal capacity, but I have been flying these types of aircraft for more than 20 years, with over 8,000 hours, and have reasonable familiarity with how the charter world operates. I feel compelled to post after seeing commentary about the Lear 60 crash, from missile strikes to DEI to critiquing this poor woman’s final radio calls.
That last part is what I find most troubling. She was the second-in-command (SIC), and while she was empowered to point out mistakes as part of good crew resource management, she was doing her job, communicating with ATC right to the end in a terrifying situation. She likely knew before many of those calls that she was going to die. I have heard final calls from friends, from military pilots with thousands of hours in fighter aircraft, and they are just as heartbreaking because they know the end is approaching.
I understand the public interest given the footage and audio. But you do not know the aircraft’s maintenance history, whether there was a structural failure, if any failure was induced by turbulence, overspeed, or overcontrol, or everything happening in that cockpit. I am sure weather will be part of the investigation, but beyond that, it is speculation. The charter industry routinely hires commercial pilots very early in their careers to gain experience alongside an ATP-qualified captain before they move up to the airlines. She would still have the appropriate training, qualifications, and checkouts by an FAA examiner or designee.
Let people investigate, learn, and share the facts. Be respectful to those who perished and the families who carry their memories. God bless them and their loved ones.
No one seems happy at this news.
Both $MRVL (-7%) and $LITE (-11%) down at the moment.
Both are down since the news leaked last week.
This looks like the largest tech merger EVER
investor.marvell.com/news-ev…
(unless you include SpaceX/xAI, but that was related party)
2
150
The fact that any model provider can just decide they don’t want to let you use their tools for something, whenever without notifying you, is concerning.
Thank goodness open models exist.
Usually the more minutiae you add (like a debt creditor would model, or actuaries), the further you are away from a fund returner.
No one is wrong per se, just different priorities.
A lot of talk on this from both sides. The good thing about X, like in financial markets, is that it doesn't matter who you are, what background you have or your seniority, you are either right by being closest to the fundamental truth or you are not. It is just this simple. So pretty nice to see guys like Shaun engaging in debate with others, this is one of the best things about X.
That being said, having seen this, I gotta say most money is made when you have a differing view from consensus and it materializes (at least in some way). For this, most of the time, napkin math is enough. This is pretty much the case with SpaceX today (and more so 10 years ago). It is also the case with most VCs, just don't pretend to do a detailed model. There is no intellectual superiority by having a 1500 row 10 tab model (now btw this can be done with AI in minutes) vs a guy that does some napkin math and looks at the 2-3 things that really matter. Actually, that is pretty much more valuable as you are able to sinthetize the business model into the 2-3 things that are paramount and are able to grasp the big picture and monitor that, while the rest is losing time on non-material stuff such as updating the +2000 line model and making sure there are no mistakes. The truth is you don't need a financial model a lot of the times. Btw, most financial models are BS anyway as they are horizontal and do not carry true unit economics. They are just endless tabs and rows that are not better than a 1 tab 400 row model, they are in fact worse due to the analyst being prone to error.
I see a lot of ppl, specially in quant finance or some pods, thinking that models ARE reality. They end up spending so much time on the model and adding so much stuff that they just stop looking at reality and just look at the models over and over, changing them, toying with them, updating them, thinking they are the actual reality. Models are abstractions and simplifications of reality, so you need to ALWAYS be aware of this. As @elonmusk says, "reality is the ultimate update", so you better have this burned in your brain. That being said, models are really useful as toy models that help you understand what are the 2-3 things that matter, what are the things that make the company gain or lose significant value... and so on. Keep it as simple as possible. The hard part is having a differing view on those 2-3 things that matter and being correct! The ppl that really make a lot of money are able to focus on these and the big picture and end up being correct when consensus believed otherwise. Remember that.
2
52
Spent a couple days sizing creative AI. The range came out $9B to $300B.
Even the bear case, where AI-natives just take share of today's spend, is 2-4x their combined run rate today.
It all hinges on one question: does cheap creation grow the market, or just reshuffle it?
Public markets are pricing a fixed pie. Private markets are pricing it doubling. Both can be right, and the incumbents still lose share.
Tell me where I’m wrong.
5
101
i'm pleasantly surprised to hear, that's awesome.
but 24 months is a long time. too long for this to fade into the ether.
Cisco gives 27-year-old engineer unlimited AI tokens to rewrite 98% of the company's legacy code in 24 months — president plucks former startup founder, telling him to 'go nuts' tomshardware.com/tech-indust…
63
Lots of fervor and noise in creative AI lately. Huge rounds and growth, while the incumbents are down.
I got curious, went back to past and current examples (Uber, App Store), and tried to sanely frame it.
xitter.gq/stephenschou/status/21…
1
4
172