At a16z investing in AI & Infra. 2x founder & CEO. CTO at Intel & VMware. CPO at Yubico. Tweets are my own.

Joined January 2008
370 Photos and videos
Pinned Post
Super excited to announce that we are leading the Series A in @netris_io. Every technology cycle changes the underlying infrastructure. And Netris has build the Networking Software Stack for AI. We are excited to partner with @alex_saroyan, Arsen Arakelyan and Tigran Martirosya! a16z.com/announcement/invest…
2
3
24
2,928
Guido Appenzeller reposted
HT @sidravi_
32
89
1,334
54,947
RT @alex_saroyan: In 2018, @tmartiro , @ars_arakelyan , and I started @netris_io . The hyperscalers had proven what's possible when you aut…
1
131
Guido Appenzeller reposted
Netris has raised its $15 million Series A, led by @a16z. Guido Appenzeller (@appenz) led the round and is joining our board, with fellow a16z partners Martin Casado (@martin_casado) and Raghu Raghuram (@RaghuRaghuram) behind the investment. Together, they're the team behind the network virtualization that reshaped the modern data center — Nicira, NSX, VMware. They solved networking for the data center. Netris is doing it for AI. Eight years ago, that bet looked wrong. We believed the network would become the hardest problem in computing, and that nobody outside the hyperscalers had the tools to solve it. The hyperscalers had built that automation for themselves — armies of engineers, custom systems. Everyone else was stuck. So we built a platform to put it within reach of any operator. Most investors didn't see it. Most of the industry didn't either. But the operators actually running networks did — they started choosing Netris long before AI was on anyone's roadmap. Then the AI buildout hit, and the network became the bottleneck for the entire industry. A single GPU cluster runs across multiple network fabrics at once — Ethernet, InfiniBand, NVL72 — each with its own control plane, spanning hundreds or thousands of switches. None of it was built to work as one, and the automation tools that came before were never designed for it. One misconfiguration can take an entire cluster down. It was the exact problem we'd spent years solving — and the platform was already built for it. Netris automates network operations across every fabric and delivers true multi-tenancy through hard isolation (enforced on networking hardware). Operators can now run shared GPU clouds, maximize GPU utilization, and provision new tenants instantly. In the last 12 months, Netris experienced 800% ARR growth and reached 35 live deployments and growing — more live deployments than all other network automation vendors combined. The largest AI clouds in the world have chosen Netris: neoclouds, sovereign AI operators, and AI factories. Huge thanks to the Netris team, our customers, investors, @nvidia, and our ecosystem partners, including @MirantisIT, @rafaysystemsinc, @RedHat, @spectrocloudinc, @vclusterlabs, and @HPE. The AI buildout is just beginning. Netris has been ready for this for eight years. Read the press release: netris.io/blog/andreessen-ho…
3
4
17
1,180
Congratulations Matic Team!
Congrats to the Production team on building 10k Matics!
2
9
767
Guido Appenzeller reposted
Today, we launch Mirelo SFX Plugin for @Roblox You can now generate scene-aware, custom sound effects like collisions within seconds based on the context of your layout and objects #RobloxDev #robloxdevs #gamedev #Roblox
3
5
22
1,661
This is kind of cool. Amazing how Sandboxes have evolved from a mini-VM to an management primitive for agents and tools.
1
1
11
974
This is misunderstanding AI. It’s an infrastructure technology that will be used by everyone. When the CPU (in the 70s) or the internet (in the early 90s) proliferated, you could have written the same about home computer use or early web browsers. How it’s a choice, and some will use it some won’t. But that’s not what happened. Infrastructure technologies become ubiquitous. Everyone uses CPUs and networking today. The may just no longer notice.
No, everyone is not using AI for everything. gabrielweinberg.com/p/people… People are consuming AI like they eat meat: some are embracing it, some are limiting their use of it, and some are avoiding it altogether.
4
807
It's silly how the @arena and @ArtificialAnlys leaderboards for open-weights image generation differ right now. - @ideogram_ai is #1 vs. #8 (really?) - @nvidia Cosmos is #8 vs. #1 - @bfl_ai is #3 vs #6 I may be biassed, but @arena seems a lot closer to reality.
14
2,032
Guido Appenzeller reposted
Flash-KMeans was only the beginning. Today, from the Flash-KMeans team, we are releasing FlashLib — a GPU library for fast, predictable, agent-ready classical ML operators. Up to 26× on KMeans, 19× on KNN, 40× on HDBSCAN, 208× on TruncatedSVD, 47× on PCA, 147× on exact t-SNE, and 49× on MultinomialNB over state-of-the-art (cuML). Blog: flashml-org.github.io/ Code: github.com/FlashML-org/flash…
47
232
1,602
868,323
Guido Appenzeller reposted
Welcome to the team, Han Wang! Han joins as VP of Engineering, bringing expertise in computational mathematics, national security, and cybersecurity, with hands-on experience leading large-scale engineering infrastructure from Lawrence Livermore National Laboratory to the U.S. Army Reserve Cyber Corps. Han also co-founded YC-backed Upfort and most recently served as Head of Engineering, Infrastructure at Numeral. His expertise comes at a pivotal moment: as we build the infrastructure to train and scale our foundation models like Pearl, having the right engineering leadership in place is critical. Learn more and explore our open roles: genesis.ml/careers
4
5
2,912
If you happen to be at CVPR, we are hosting a happy hour. Expect food, drinks, but most importantly super interesting people.
Happy hour at #CVPR2026 tomorrow 🍸 Come meet @a16z infra a dozen of the best generative media & computer vision teams in one place. I can promise great food, drinks, and company! RSVP👇
3
553
TL;DR: a16z is putting more emphasis on international
4
489
A great new meaningful benchmark for molecular models is out, and @genesismolai's Pearl model is doing extremely well. It looks like scaling laws work for protein folding just like they do for language models, image generation or video.
Today we're sharing new breakthrough results for Pearl, our foundation model for protein–ligand cofolding. The OpenBind Consortium recently released the first public structure-affinity benchmark for molecular AI, evaluating six prominent cofolding models on the EV-A71 2A protease. We ran our full Pearl system against the same target. Zero-shot, with no binding-site information and no tuning, the Pearl system reaches 78% on OpenBind's primary success criteria, far ahead of every cofolding model tested by OpenBind. We also assessed a stricter sub-1 Å accuracy threshold, which is more relevant for real-world R&D usage – the Pearl system’s success is still 60%, versus 1–27% for the other models. What matters most to us: this is the same system setup our scientists use on live drug discovery programs, not a benchmark-specific configuration. Thanks to the OpenBind Consortium for building a rigorous public benchmark, and to @NVIDIAHealth for the support on optimizations that enabled model scaling.
2
587
Wohoo! Great to see Ideogram launching an open weights model!
Introducing Ideogram 4.0: the best open image model in the world. Think it. Make it. Own it. Download the weights, fine-tune on your own data, and run it on your hardware. Live on every Ideogram plan and the API today.
8
452
If you happen to be at CVPR, give us a shout. Personally I’ll only be there Friday, but our team is in Denver the whole time.
Going to CVPR in Denver this week? Our team at @a16z and several of our portfolio companies are hosting events! Come hang with us - links 👇
2
409
Guido Appenzeller reposted
Announcing @a16z's investment in Town 🎉 It's an AI assistant that truly knows you. Since I joined Town a few months ago, it's run my work life - emails, to-do lists, meeting prep, reminders. It can actually DO things for you and has full context. x.com/jgreze/status/20621786…
24
11
220
34,948
Whoaaa… huge congratulations to @BenMildenhall and @_pratul_, amazing accomplishment!
Congrats @BenMildenhall and @_pratul_! ACM gives Grace Hopper Awards to under-35-year-olds for *one specific thing*, here it's NeRF. Past winners include Donald Knuth (for "The Art of Computer Programming"), Steve Wozniak (for Apple), and @martin_casado (for SDN). Good company!
4
885
Never trust Google's AI answers for laws and regulation. Here it shows the original text in German and in the English translation adds an extra term (the word "proven"). It also told me a PC-12 can't land there as multi-engine planes need certification (PC-12 is single engine).
1
333