tweetindex
PT

Patrick C Toulme ✓

@PatrickToulme · New York, NY · joined 27 Dec 2025

gemini & TPUs @google. Formerly worked on Meta MTIA and AWS Trainium. Opinions are my own. All blogs shared are personal work.

13 494Followers
386Following
826Posts total
1.2MViews on collected posts

Posts recentes

$META My hypothesis for how Muse Spark 1.3 is so cheap - they are partially serving it on MTIA chips. This would allow them to serve the model very cheaply and efficiently, and explains where they got all the compute from to suddenly vend a frontier model from. This would mean 13.5K views · 125 likes · 9 reposts · 12 replies Open on X →
@PatrickToulme if watermelon is truly comparable to fable as theorized then meta acting in their own interest hurts anthropic as a mere second order effect. i don't even see ill intent here tbh 942 views · 6 likes · 0 reposts · 0 replies Open on X →
@PatrickToulme This just proves it is bruteforce rather than intelligence 7.3K views · 25 likes · 1 reposts · 4 replies Open on X →
@PatrickToulme 🤝 same page 1.1K views · 13 likes · 0 reposts · 1 replies Open on X →
@PatrickToulme Wishful thinking. Nobody is turning down Claude's use. The model is just way too good. 1.4K views · 5 likes · 0 reposts · 0 replies Open on X →
@PatrickToulme Does $META have enough compute to put all their employees on Watermelon? Most of them are in the same time zone so utilization rate won’t be efficient either. 10.9K views · 35 likes · 0 reposts · 5 replies Open on X →
@PatrickToulme OpenAi’s customers could do the same. So could any companies customers. 1.9K views · 2 likes · 0 reposts · 0 replies Open on X →
$META I predict will imminently shutoff Anthropic and OpenAI model access for employees. This may very well affect the Anthropic and OpenAI IPOs. Watermelon will be Fable class. This is 100% proof that Meta's experiment of having employees generate tens of thousands of RL envs 476.6K views · 3K likes · 131 reposts · 76 replies Open on X →
Meta could shutoff Claude access for employees, and you would see Anthropic revenue tank. Meta might very well do this to mess up the IPO, especially if watermelon is Fable quality. 125.2K views · 462 likes · 22 reposts · 23 replies Open on X →
New from Ramp data: the latest threat to the AI trade. AI companies' revenues are heavily dependent on a small set of customers. 80% of OpenAI and Anthropic's enterprise revenues come from 1% of their customers, and it's not getting better. This is a level of concentration risk 539.5K views · 1.3K likes · 175 reposts · 135 replies Open on X →
The more I think about $NVDA situation the more I believe RSI is crucial for their future. CUDA is rapidly becoming irrelevant in frontier AI work. My thesis: CUDA is a software moat guarding Nvidia silicon. CUDA was a significant moat when humans wrote all kernels and software 21.7K views · 143 likes · 6 reposts · 16 replies Open on X →
OpenAI shutting off GPT access to @cursor_ai and @SpaceXAI is 100% because they see xAI and Grok as a legitimate competitor. This is actually a very big compliment to Grok. In my view - Grok has fully matched Opus in capability and xAI will imminently have a Fable quality 3.6K views · 39 likes · 2 reposts · 5 replies Open on X →
This is Jevon’s Paradox in action. Cost per token will go close to zero 4.3K views · 16 likes · 1 reposts · 5 replies Open on X →

Em comparação com contas do mesmo porte

13 posts dos últimos 90 dias, ao lado da faixa de 10K–100K seguidores. aparece para muita gente, mas poucos desses espectadores reagem.

Mediana de visualizações7 336esta conta924mediana para 10K–100K
Alcance, %54.36%esta conta3.62%mediana para 10K–100K
Engajamento, %0.51%esta conta1.52%mediana para 10K–100K
MétricaEsta contaMediana para 10K–100KProporção
Mediana de visualizações por post7 3369247.94×
Alcance (visualizações ÷ seguidores)54.36%3.62%15.0×
Taxa de engajamento0.51%1.52%0.33×

Outras contas desta faixa →   Comparar com outra conta →   Como estas referências são construídas →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

4.3K29 Aug
3.6K
21.7K1 Sep
539.5K2 Sep
125.2K
476.6K
1.9K
10.9K
1.4K
1.1K
7.3K
942
13.5K3 Sep

Last 13 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.51%29 Aug
1.27%
0.78%1 Sep
0.32%2 Sep
0.41%
0.68%
0.11%
0.37%
0.37%
1.27%
0.41%
0.64%
1.10%3 Sep

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.

What the audience does

Likes66.0%5 151 in total
Reposts4.4%347 in total
Replies3.6%282 in total
Quotes2.0%155 in total
Bookmarks24.0%1 872 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Contas semelhantes

Frequently asked questions about @PatrickToulme

How many followers does @PatrickToulme have?

13 494 followers.

How many posts has @PatrickToulme published?

826 posts.

How do posts by @PatrickToulme perform?

Collected posts: 13. Views: 1 207 986. Likes: 5 151.