Sander Arts

11K posts

Sander Arts banner
Sander Arts

Sander Arts

@Sander1Arts

CMO | Advisor | Board Member | Author | People to People Marketing

Katılım Temmuz 2010
3.4K Takip Edilen3.6K Takipçiler
Sander Arts retweetledi
Aria Networks, Inc.
Aria Networks, Inc.@AriaNetworks·
Next week at @DataCentre_Cong North America — @AriaNetworks' Chris Summers is presenting: "The Network: The Key Lever in Improving AI Factory MFU and Token Efficiency" The network is 10–15% of cluster spend. A 1% MFU gain offsets the entire investment. It's not plumbing. It's the highest-leverage variable in your AI factory. 👇 Register datacentrecongress.com/northamerica/
English
1
4
8
230
Aria Networks, Inc.
Aria Networks, Inc.@AriaNetworks·
The network is no longer plumbing. As distributed inference, reasoning models, and agentic AI systems scale, it becomes one of the highest-leverage components in the entire AI factory — and one of the least understood. Grateful to Shane Snider and @datacenter for the opportunity to dig into this thinking in a recent feature. datacenterknowledge.com/infrastructure… @mansourkaram
English
1
2
6
259
Sander Arts retweetledi
Mansour Karam
Mansour Karam@mansourkaram·
Inference is becoming one of the hardest systems problems in AI infrastructure. For a long time, the mental model was relatively simple: a request goes to compute, the model runs, and a response comes back. The new reality is distributed inference: reasoning models, agents, KV cache movement, prefill/decode disaggregation, memory pressure, storage interaction, and bursty traffic patterns that can affect thousands of users at once. In that world, the network is not plumbing. It is one of the highest-leverage systems in the AI factory. At @AriaNetworks , we believe the future of AI networking will be measured not only in latency and throughput, but in MFU, token efficiency, and cost per token. I appreciated the conversation with Shane Snider at @datacenter on why inference is reshaping data center networking. #AIInfrastructure #Inference #DeepNetworking #Ethernet datacenterknowledge.com/infrastructure…
English
0
4
12
441
Mike Investing
Mike Investing@MrMikeInvesting·
We are currently in a “once in a lifetime” AI super cycle… Phase 1 was: (already gone) Semiconductors ~ $NVDA, $AMD, $INTC, $ARM Phase 2 is: (passing by now) Memory ~ $MU, $SNDK, $WDC Photonics ~ $AAOI, $AEHR, $LITE, $MRVL The current phase is Neo Cloud/AI infrastructure: $IREN, $NBIS, $CRWV, $CIFR, $APLD Next wave (many will miss) Rare Earths ~ $USAR, $MP, $UUUU, $FCX Power & Cooling~ $VRT, $CEG, $OKLO, $OSS Finally it all concludes with these 3 sectors: Robotics ~ $TSLA, $PATH, $SERV Space ~ $RKLB, $ASTS, $PL, $LUNR Drones ~ $ONDS, $AVAV, $LMT Many will make generational wealth from this AI super cycle over the next 7 months. Save this to look back on later…
English
162
747
7.2K
750.2K
Sander Arts retweetledi
Aria Networks, Inc.
Aria Networks, Inc.@AriaNetworks·
In the 1920s, factories finally got electric motors and bolted them straight where the steam engine used to be. Productivity barely moved. For thirty years. That's what most AI infrastructure vendors are doing today. Same architecture. Same dashboards. Just with a chat window on top. In his latest blog, @djspry breaks down the three frameworks shaping how @AriaNetworks decides what to build and why the context layer is where durable value actually accumulates. Worth a read if you're building anything in the AI era. 🔗 arianetworks.com/news/old-patte…
English
0
2
5
165
Sander Arts retweetledi
Aria Networks, Inc.
Aria Networks, Inc.@AriaNetworks·
May the 4th be with you. 📷 In a galaxy of AI factories, the network is everything. Every Jedi knows: raw power means nothing if it's poorly directed. The same is true for AI infrastructure. You can pack a data center with the most powerful accelerators in the universe, but if your network is congested, every idle GPU, every stalled training job, every microsecond of latency is a direct tax on your outcome. The Force multiplier isn't the hardware. It's the network. At Aria, we built the networking platform for the AI factory era from scratch. Not adapted from legacy enterprise infrastructure, not a bolt-on. Designed to convert accelerators into intelligence, and intelligence into revenue. Maximum Goodput. Real-time telemetry at 100x the resolution of traditional tools. Open, vendor-agnostic, and in production today. May your GPUs be fully utilized, your congestion windows clear, and your MFU forever in your favor. @mansourkaram
English
0
1
3
121
Sander Arts
Sander Arts@Sander1Arts·
Love working with @mansourkaram + team.
Aria Networks, Inc.@AriaNetworks

The launch. The demo. The Network That Thinks. Live and on record. Our YouTube channel is now live with everything from the GA launch. The full event. The live Deep Networking demo. The people who showed up and said things out loud that the industry has been thinking quietly. And this is just the beginning. We will be adding more content regularly as we continue to build, ship, and show our work. All of it. One place. Come see. @Aria_Networks" target="_blank" rel="nofollow noopener">youtube.com/@Aria_Networks

English
0
0
0
26
Sander Arts retweetledi
Mansour Karam
Mansour Karam@mansourkaram·
This is what we see every day at @AriaNetworks, and the problem #DeepNetworking was built to solve. The AI infrastructure conversation is focused on xPUs and power. But the network governs how quickly gradients synchronize, how efficiently KV caches transfer, how seamlessly jobs schedule across thousands of xPUs. When it underperforms, it drags down every other component in the stack. As @DellOroGroup put it: "The network is no longer just plumbing — it's a first-order limiter of AI performance." Exactly right. You can't solve a network problem by buying more xPUs.
Data Center Knowledge@datacenter

🚨 The Breaking Points: Networking Strains Under AI’s Scale Demands In the first of a three-part series, we explore how AI infrastructure struggles to scale, focusing on the networking challenges reshaping enterprise systems. datacenterknowledge.com/networking/the…

English
1
5
14
1.2K
Sander Arts retweetledi
The New Stack
The New Stack@thenewstack·
Aria Networks launches "Network that Thinks" initiative to optimize AI cluster performance through Model Flop Utilization, SONiC, telemetry, and AI agents. thenewstack.io/aria-networks-…
English
0
3
2
463
Sander Arts retweetledi
Mansour Karam
Mansour Karam@mansourkaram·
"Networking has traditionally been in the background, like plumbing. But in an AI-driven enterprise, the network is more like the nervous system." - Lumen's CEO. She's right. Omdia also found that many neoclouds "have scaled up their compute infrastructure to handle AI workloads, but their networking infrastructure is becoming a critical constraint." The right network determines how much value you extract from your AI investments. That's the problem we are solving with @AriaNetworks #DeepNetworking theregister.com/2026/04/15/net…
English
0
5
8
647
Sander Arts retweetledi
Aria Networks, Inc.
Aria Networks, Inc.@AriaNetworks·
Appreciate the support, @GavinSBaker. Traditional networking wasn’t built for AI - deep networking is. Big gains in MFU and tokens/sec ahead.
Gavin Baker@GavinSBaker

Delighted that @atreidesmgmt and our partners at Valor led the series A in @AriaNetworks If you are using traditional data center networking for your front-end or scale-out networking in any AI cluster, please take a look at Aria and their “deep networking” technology. You will likely get higher MFU during training and more tokens per second per watt for inference. Lowering cost per token is great, but given that tokens are now literally revenue, Aria should increase your effective revenue per GPU by driving more tokens per second. Also one of the few companies @atreidesmgmt has ever invested in that should materially beat their year 1 forecast in the weeks after we invested! @MansourKaram is exceptional imo: a successful, repeat, deep tech founder having founded and sold Apstra to Juniper. Apstra is now essentially the foundation of Juniper's entire networking tech-stack and featured prominently in all their earnings call prior to HPE acquiring them.

English
0
4
7
1K
Sander Arts retweetledi
Gavin Baker
Gavin Baker@GavinSBaker·
Delighted that @atreidesmgmt and our partners at Valor led the series A in @AriaNetworks If you are using traditional data center networking for your front-end or scale-out networking in any AI cluster, please take a look at Aria and their “deep networking” technology. You will likely get higher MFU during training and more tokens per second per watt for inference. Lowering cost per token is great, but given that tokens are now literally revenue, Aria should increase your effective revenue per GPU by driving more tokens per second. Also one of the few companies @atreidesmgmt has ever invested in that should materially beat their year 1 forecast in the weeks after we invested! @MansourKaram is exceptional imo: a successful, repeat, deep tech founder having founded and sold Apstra to Juniper. Apstra is now essentially the foundation of Juniper's entire networking tech-stack and featured prominently in all their earnings call prior to HPE acquiring them.
English
14
14
204
60.7K

Keşfet

@mansourkaram @andrewdfeldman @cerebras @djspry @DataCentre_Cong @AriaNetworks @datacenter @stefandyckerhof