Mascobot Profile Banner
Marco Mascorro Profile
Marco Mascorro

@Mascobot

Followers
16K
Following
13K
Media
748
Statuses
6K

Partner @a16z (investor in @cursor_ai, @thinkymachines, @bfl_ml, @WaveFormsAI & more) | Roboticist | Cofounder @Fellow_AI | @MIT 35 under 35 | Opinions my own.

San Francisco, CA
Joined October 2009
Don't wanna be here? Send us removal request.
@Mascobot
Marco Mascorro
11 days
🚨 New: We built @a16z's personal GPU AI Workstation Founders Edition. - 4x NVIDIA RTX 6000 PRO Blackwell Max-Q (384GB total VRAM).- 8TB of NVMe PCIe 5.0 storage.- AMD Threadripper PRO 7975WX (32 cores, 64 threads).- 256GB ECC DDR5 RAM.- 1650Watts at peak (runs on a standard
Tweet media one
545
456
5K
@Mascobot
Marco Mascorro
15 hours
Need to step up my cable management skills with my GPU servers 😍.
@elonmusk
Elon Musk
1 month
Cable pr0n of @xAI GB200 servers at Colossus 2
Tweet media one
Tweet media two
1
0
5
@grok
Grok
21 days
Turn old photos into videos and see friends and family come to life. Try Grok Imagine, free for a limited time.
729
1K
5K
@Mascobot
Marco Mascorro
3 days
Opportunity cost is so high right now, never seen anything like this….
30
29
461
@Mascobot
Marco Mascorro
6 days
I mean you could just walk there and buy those SMD resistors, robot arm actuators, connectors, etc. Akihabara in Japan has a version of this too:
Tweet media one
Tweet media two
4
1
40
@Mascobot
Marco Mascorro
6 days
San Francisco Centre (a dead mall today) could be our SF Huaqiangbei (Shenzhen electronics market) 😀
Tweet media one
Tweet media two
42
11
267
@Mascobot
Marco Mascorro
11 days
The liquid-cooled AMD Threadripper PRO 7975WX features 32 cores and 64 threads on 5nm Zen 4 (Storm Peak), with DDR5, 128 PCIe 5.0 lanes, 128 MB L3 cache, and clock speeds up to 5.3 GHz boost / 4.0 GHz base. The workstation also integrates an AST2600 Baseboard Management
Tweet media one
9
10
163
@Mascobot
Marco Mascorro
11 days
We designed this AI Workstation to deliver maximum FLOPS for AI research, in a compact, under-desk form factor, movable and accessible, just like a gaming PC.
Tweet media one
3
2
127
@Mascobot
Marco Mascorro
11 days
The RTX 6000 PRO Blackwell Max-Q delivers 24,064 CUDA cores, 96GB of GDDR7 VRAM memory, PCIe 5.0 x16, and next-gen Blackwell architecture in a remarkably efficient 300W powerhouse. The workstation integrates four of them on PCIe 5.0 x16 and 8TB NVMe PCIe 5.0 storage, unlocking
Tweet media one
8
5
171
@Mascobot
Marco Mascorro
17 days
Full episode: .
1
0
9
@Mascobot
Marco Mascorro
17 days
Robotics simulation is one of the potential Genie 3 applications. While simulators like MuJoCo are great, they still lack the real world experience. Genie 3 closes the gap between the sim2real by taking the best of both worlds: taking a real-world data driven approach with the
1
0
15
@Mascobot
Marco Mascorro
17 days
.@jparkerholder's first test of the viral painting the wall clip, including its first-person perspective, and the improvement in time of the spatial memory, going from a few seconds of Genie 2, to a 1+ min in Genie 3:
1
0
5
@Mascobot
Marco Mascorro
17 days
It's interactivity, spatial memory and consistency are some of the features that stand out. @jparkerholder explains how this was a big improvement and an effort to combine the capabilities from Genie 2, Veo 2 and the DOOM game engine paper:
1
0
5
@Mascobot
Marco Mascorro
17 days
🚨 New: Interviewed Genie 3 co-leads @jparkerholder and @shlomifruchter from @GoogleDeepMind to talk about their new interactive real-time world model, a next evolution from Veo 3, Genie 2 and the DOOM game engine paper. Genie 3 showed the world what's possible with real-time
8
44
238
@Mascobot
Marco Mascorro
19 days
RT @ericzakariasson: sending these out to a bunch of our top tab users
Tweet media one
0
22
0
@Mascobot
Marco Mascorro
21 days
I’m seeing a lot of friends too with great networks start new funds that are doing very well (invested in great AI companies early), set up SPVs, etc, most of them with great access to the best founders. It’s a great time to be doing this too.
@leopoldasch
Leopold Aschenbrenner
21 days
1
0
12
@Mascobot
Marco Mascorro
25 days
Robotics engineers: "let's design a 7 DOF robot arm with a computer vision model running at 60 fps on the edge to recognize the position of the bottles and flip them." . Vibe engineers: ". violently slap the bottles until they turn upside down".
@MachinePix
MachinePix
26 days
Rotary bottle orientor.
4
4
44
@Mascobot
Marco Mascorro
25 days
SOOO excited about Computer Use/chatGPT Agent/Operator and all the ML, VLMs, RL, infra and tooling needed for it. This space seems so new and early with so many things yet to be built.
0
1
12
@Mascobot
Marco Mascorro
26 days
Woohoo! Cursor is now in terminal:.
@cursor_ai
Cursor
26 days
Cursor is now in your terminal! It’s an early beta. Access all models. Move easily between your CLI and editor.
Tweet media one
0
0
7
@Mascobot
Marco Mascorro
26 days
GPT-5 scores 65.7% on ARC AGI 1, at only $0.51 per task:.
@arcprize
ARC Prize
26 days
GPT-5 on ARC-AGI Semi Private Eval. GPT-5.* ARC-AGI-1: 65.7%, $0.51/task.* ARC-AGI-2: 9.9%, $0.73/task. GPT-5 Mini.* ARC-AGI-1: 54.3%, $0.12/task.* ARC-AGI-2: 4.4%, $0.20/task. GPT-5 Nano.* ARC-AGI-1: 16.5%, $0.03/task.* ARC-AGI-2: 2.5%, $0.03/task
Tweet media one
0
0
2