russ (@russty2222) header
russ (@russty2222) profile picture
Joined January 2025
Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)Photo by russ (@russty2222)
2/3 4. Communication Here I have more sympathy with retail. WYFI can definitely improve communication and expectation-setting as that is largely within management’s control. But there something to be mindful of, in that a public company cannot live-stream lender negotiations, customer disputes, contractual negotiations, financing discussions and potentially material developments simply because shareholders on X want reassurance. Sometimes management genuinely is working behind the scenes while investors see nothing. I’ve experienced that from the other side of the boardroom table. As an ex-auditor I’ve worked some pretty crazy hours thinking my job was tough, while businesses were firefighting what felt like multiple dumpster fires simultaneously. From outside, someone might see one missed target or wonder why management hadn’t communicated something. Inside the boardroom, I could see the CEO of one well known Indian Steel conglomerate dealing with 10-+ interlinked problems all at once, from financing, customers, contracts, supply chain, systems, litigation, employees, tax authorities, deadlines and auditors on a quarterly basis too 🤯 Most of these sort of things cannot be discussed publicly. I’ve even had one CEO tell me horror stories that he was dealing with and he hadn’t seen his family with young kids for over 2 weeks… Seeing both sides changed me as an investor. I have more patience now. Not blind patience though. If management controls something, hold them accountable for it. If they promise something, track it. If financing terms are poor, analyse them. If contracts don’t turn into revenue, ask why. If communication can be better, say so. But equally, don’t confuse something being outside management’s control with management incompetence. And don’t confuse “Management hasn’t told retail what is happening” with “Nothing is happening.” Especially when the observable evidence says the opposite and that quite a lot has been happening for a company this young. What you ask? Well, since IPO: 1. WYFI has moved NC-1 into customer deployment and signed the ~40MW / ~$865M Nscale agreement. That’s billing has already commenced… 2. They’ve accumulated hundreds of millions of dollars of additional Cloud contract value with several clients old and new, got VR thrown in the mix too. 3. Agreed terms for NC-2/NC-3 acquisition, have LoIs signed for these sites. 4. Acquired MTL-3, and resumed construction on MTL-2 with near term completion target and tenant. They’ve also applied for expansion on MTL-1 and MTL-3 so clearly demand for what is produced is increasing and WYFI is making moves to supply. 5. Partnered with Krambu around future high-density infrastructure capacity and R&D 6. Commercially launched Continuum and continuing to build out its capability. There’s more they’ve done which I have forgotten, but all this was done while responsibly raising the capital required to build an extraordinarily capital-intensive business. It doesn’t mean every decision has been perfect. But the record should be judged as a whole. …continued👇🏼…
1
2
1
11
1,100
Presented by our partners at @WhiteFiber_ $WYFI: Founder and CEO Sam V. Tabar joins Boardroom Exclusives to talk about retrofitting abandoned industrial facilities into data centers, driving momentum in their GPU cloud business, and launching a new technology that aggregates power sites into a virtual supercluster optimized for AI inference.
5
15
3
431
3,471,937
🚨 BIGGEST MISCONCEPTION IN DATA CENTER PUSHBACK: Water. Closed-loop systems + dry coolers recirculate the same fluid. Heat is rejected without using towers that draw millions of gallons. There is very little use on municipal water supply. $WYFI is definitely one of the most responsible data center developers. The sites are also former industrial plants with electrical infrastructure already in place.
7
9
1
72
7,684
Responsible data center development is a priority for WhiteFiber. Billy talks about what this means and why we don't just enter communities, we become part of them. $WYFI #datacenters
7
16
7
101
28,905
Congratulation to Michael, Toby and all the @WhiteFiber_ team. Truly eye opening patented Tech. All that hard work coming to fruition $WYFI - 136 Tbps bandwidth, 0.9 millisecond round-trip latency over 83 KM site-to-site. 🤩 So What’s Next • Two sites isn’t the ceiling. Continuum’s Jericho3-AI fabric can already run a single cluster of 32,000 GPUs at 800Gbps—about 62× what ATL1 and SUW1 run today. The bottleneck isn’t the network. It’s how many sites and how much power you plug in. • New GPUs, same fabric. Continuum is on H200s because that’s what’s shipping. When the next NVIDIA generation lands, DriveNets Ethernet and the same NIC choices (ConnectX, Broadcom, others) carry over. No redesign. • Jericho4 is the next leap. Deep buffering + congestion control for lossless RoCE past 100 km, plus line-rate MACsec on every port. Not in Continuum yet—but it’s the near-term path to longer-distance AI and a stronger security story for regulated customers.
Replying to @WhiteFiber_
@WhiteFiber_ Continuum is live. Built with @drivenets and @weka, it's the highest-performing distributed GPU supercluster architecture. Bandwidth is no longer a design constraint. Read the blog: whitefiber.com/blog/continuu… #AIInfrastructure #WhiteFiberContinuum
2
7
29
2,184
Replying to @WhiteFiber_
@WhiteFiber_ Continuum is live. Built with @drivenets and @weka, it's the highest-performing distributed GPU supercluster architecture. Bandwidth is no longer a design constraint. Read the blog: whitefiber.com/blog/continuu… #AIInfrastructure #WhiteFiberContinuum
1
11
506
🚨 @WhiteFiber_ Launches Continuum: First Commercial Distributed GPU Supercluster 🚨 ⚡ 136 Tbps bandwidth, 0.9ms latency — connecting two HITRUST-certified facilities 83km apart over 12 Zayo dark fiber strands into a single logical GPU supercluster ⚡ Patent applications filed — for the underlying architecture, built alongside @drivenets (AI network fabric) and @weka (NeuralMesh storage/memory) ⚡ Turns fragmented power into scale — lets metro/telco sites with stranded power and fiber capacity contribute to one pooled supercluster instead of sitting underutilized in isolation ⚡ Solves data sovereignty for regulated industries — sensitive workloads stay within their originating jurisdiction even while failover and pooled compute happen across sites ⚡ No single point of failure — training runs continue even if one location goes down, rather than stalling until it's restored $WYFI
2
1
30
1,867
First of all, congratulations to Michael and the entire @WhiteFiber_ team. They have officially launched Project Redwood under its commercial name, WhiteFiber Continuum, delivering 136 TB per second of aggregate bandwidth and 0.9 milliseconds of round-trip latency across 83 kilometres. These are very impressive specifications. We spoke with management about WhiteFiber’s strategy and build-out, and I want to explain the potential impact of this solution on the addressable supply of AI-ready power. I recently shared that Anthropic and OpenAI are reportedly exploring smaller 20–30 MW deployments alongside their multi-hundred-megawatt and gigawatt campuses. We know that all these players are trying to bring compute online as quickly as possible, while not every workload needs to run at one physical location. There are naturally far more industrial sites with smaller power connections than locations offering hundreds of megawatts, or even a gigawatt, in one place. The market for acquiring buildings with these smaller connections is therefore much larger. With Continuum, WhiteFiber has, in my opinion, designed a smart solution for connecting these separate blocks and making them operate as one logical GPU environment. With 0.9 milliseconds of round-trip latency, multiple 20 MW locations could potentially be connected into a much larger logical cluster. This could be an important step in a fragmented market of smaller locations, and it fits WhiteFiber’s wider strategy perfectly: • Retrofit smaller industrial locations faster • Expand through partner facilities, with Krambu as the first announced example • Bring the cluster design, customers and operating expertise • Connect owned and partner sites through Continuum • Monetize capacity without needing to own every building itself In my opinion, this creates a very fast time-to-power model. It also gives WhiteFiber the option to connect partner locations instead of only using each site on a standalone basis. It connects WhiteFiber’s retrofit strategy, partner network, cloud platform and managed-services ambitions into one broader infrastructure model. I think its current strategy aligns very well with what the market needs today, especially with long development lead times and growing community pushback against enormous data-center campuses. I will be following this closely. You can read the call with $WYFI management in comments👇 recommend it.
Replying to @WhiteFiber_
@WhiteFiber_ Continuum is live. Built with @drivenets and @weka, it's the highest-performing distributed GPU supercluster architecture. Bandwidth is no longer a design constraint. Read the blog: whitefiber.com/blog/continuu… #AIInfrastructure #WhiteFiberContinuum
3
7
1
59
6,383
@WhiteFiber_ Continuum is live. Built with @drivenets and @weka, it's the highest-performing distributed GPU supercluster architecture. Bandwidth is no longer a design constraint. Read the blog: whitefiber.com/blog/continuu… #AIInfrastructure #WhiteFiberContinuum
12
16
10
75
15,946
JUST IN: WhiteFiber launches Continuum, a distributed GPU supercluster platform linking geographically separated data centers into a single system to enhance AI infrastructure bandwidth and throughput. $WYFI
1
2
14
685
📢 𝗝𝗨𝗦𝗧 𝗜𝗡: $WYFI WhiteFiber Launches Continuum Distributed GPU Supercluster Across Data Centers 👉 𝗞𝗲𝘆 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀: ➤ 𝗪𝗵𝗶𝘁𝗲𝗙𝗶𝗯𝗲𝗿 launches 𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝘂𝗺 for cross-data-center GPU networking. ➤ Company claims industry's first commercially available 𝗱𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗲𝗱 𝗚𝗣𝗨 𝘀𝘂𝗽𝗲𝗿𝗰𝗹𝘂𝘀𝘁𝗲𝗿 architecture. ➤ System connects two QTS data centers located 𝟴𝟯 𝗸𝗶𝗹𝗼𝗺𝗲𝘁𝗲𝗿𝘀 apart. ➤ Network uses 𝟭𝟮 𝗭𝗮𝘆𝗼 𝗱𝗮𝗿𝗸 𝗳𝗶𝗯𝗲𝗿 strands between facilities. ➤ Architecture delivers 𝟭𝟯𝟲 𝗧𝗯𝗽𝘀 bandwidth with guaranteed 𝟬.𝟵𝗺𝘀 round-trip latency. ➤ 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 provides AI network fabric for the distributed architecture. ➤ 𝗪𝗘𝗞𝗔 provides storage and memory infrastructure through its 𝗡𝗲𝘂𝗿𝗮𝗹𝗠𝗲𝘀𝗵 platform. ➤ Enterprises can pool 𝗚𝗣𝗨 resources across geographically separated data centers. ➤ Architecture targets 𝗱𝗮𝘁𝗮 𝘀𝗼𝘃𝗲𝗿𝗲𝗶𝗴𝗻𝘁𝘆, resiliency, training, and inference workloads. ➤ WhiteFiber has submitted 𝗽𝗮𝘁𝗲𝗻𝘁 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 covering the underlying implementation. 👉 𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗠𝗮𝘁𝘁𝗲𝗿𝘀: ➤ Could expand AI compute beyond the 𝗽𝗼𝘄𝗲𝗿 𝗹𝗶𝗺𝗶𝘁𝘀 of individual facilities. ➤ Enables organizations to combine 𝗚𝗣𝗨 𝗰𝗮𝗽𝗮𝗰𝗶𝘁𝘆 across separate physical locations. ➤ Geographic redundancy could help 𝗔𝗜 𝘄𝗼𝗿𝗸𝗹𝗼𝗮𝗱𝘀 continue during site outages. ➤ Opens potential AI capacity from underused 𝗽𝗼𝘄𝗲𝗿 and 𝗳𝗶𝗯𝗲𝗿 infrastructure. 👉 𝗘𝘅𝗽𝗲𝗿𝘁 𝗦𝘁𝗮𝘁𝗲𝗺𝗲𝗻𝘁: 𝗦𝗮𝗺 𝗧𝗮𝗯𝗮𝗿, Chief Executive Officer of WhiteFiber: "WhiteFiber Continuum is the infrastructure answer to a problem the industry has lived with for years: geographic distance as a ceiling on what a cluster can do."
2
1
10
833
📢 𝗝𝗨𝗦𝗧 𝗜𝗡: $WYFI WhiteFiber Launches Continuum Distributed GPU Supercluster Across Data Centers 👉 𝗞𝗲𝘆 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀: ➤ 𝗪𝗵𝗶𝘁𝗲𝗙𝗶𝗯𝗲𝗿 launches 𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝘂𝗺 for cross-data-center GPU networking. ➤ Company claims industry's first commercially available 𝗱𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗲𝗱 𝗚𝗣𝗨 𝘀𝘂𝗽𝗲𝗿𝗰𝗹𝘂𝘀𝘁𝗲𝗿 architecture. ➤ System connects two QTS data centers located 𝟴𝟯 𝗸𝗶𝗹𝗼𝗺𝗲𝘁𝗲𝗿𝘀 apart. ➤ Network uses 𝟭𝟮 𝗭𝗮𝘆𝗼 𝗱𝗮𝗿𝗸 𝗳𝗶𝗯𝗲𝗿 strands between facilities. ➤ Architecture delivers 𝟭𝟯𝟲 𝗧𝗯𝗽𝘀 bandwidth with guaranteed 𝟬.𝟵𝗺𝘀 round-trip latency. ➤ 𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 provides AI network fabric for the distributed architecture. ➤ 𝗪𝗘𝗞𝗔 provides storage and memory infrastructure through its 𝗡𝗲𝘂𝗿𝗮𝗹𝗠𝗲𝘀𝗵 platform. ➤ Enterprises can pool 𝗚𝗣𝗨 resources across geographically separated data centers. ➤ Architecture targets 𝗱𝗮𝘁𝗮 𝘀𝗼𝘃𝗲𝗿𝗲𝗶𝗴𝗻𝘁𝘆, resiliency, training, and inference workloads. ➤ WhiteFiber has submitted 𝗽𝗮𝘁𝗲𝗻𝘁 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 covering the underlying implementation. 👉 𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗠𝗮𝘁𝘁𝗲𝗿𝘀: ➤ Could expand AI compute beyond the 𝗽𝗼𝘄𝗲𝗿 𝗹𝗶𝗺𝗶𝘁𝘀 of individual facilities. ➤ Enables organizations to combine 𝗚𝗣𝗨 𝗰𝗮𝗽𝗮𝗰𝗶𝘁𝘆 across separate physical locations. ➤ Geographic redundancy could help 𝗔𝗜 𝘄𝗼𝗿𝗸𝗹𝗼𝗮𝗱𝘀 continue during site outages. ➤ Opens potential AI capacity from underused 𝗽𝗼𝘄𝗲𝗿 and 𝗳𝗶𝗯𝗲𝗿 infrastructure. 👉 𝗘𝘅𝗽𝗲𝗿𝘁 𝗦𝘁𝗮𝘁𝗲𝗺𝗲𝗻𝘁: 𝗦𝗮𝗺 𝗧𝗮𝗯𝗮𝗿, Chief Executive Officer of WhiteFiber: "WhiteFiber Continuum is the infrastructure answer to a problem the industry has lived with for years: geographic distance as a ceiling on what a cluster can do."
4
5
6
43
9,143
$WYFI WhiteFibers Five Moats that nobody is talking about #1 Colocation “Retrofit” EX abandoned industrial buildings into Tier 3 AI Data centers (local communities much prefer Retrofits while getting the same property tax benefits) built 2X faster and 40% cheaper #2 WhiteFibers own Cloud - WhiteFiber owns the building & GPU’s #3 Asset light Cloud - Partner/Tenant owns the building & WhiteFiber owns/install the GPU’s #4 Managed Services - Partner/Tenant owns the building & GPU’s and WhiteFiber designs/installs the GPU’s & Manages the Cloud # 5 Redwood “Continuum” - WhiteFiber sells their patented tech to Hyperscalers, NeoClouds and Enterprises to “connect” smaller Data Centers as one (within 83KM) - Ger Ready @WhiteFiber_ for a crazy few months
Something's coming September 23. Stay tuned. #AIInfrastructure #GPUCloud
3
4
1
38
2,626
russ retweeted
$WYFI very own Luna Tan has done it again… Meeting $NVDA team including VPs on all things AI. Over a 4 month period, $WYFI signed deals worth $540m TCV. “Vera Rubin on the horizon” - more to come? First 576-GPU VR contract w/ @PrimeIntellect scheduled Q2 2027 ~ $108m TCV
$WYFI Head of Operations with AI supreme leader Jensen Huang at GTC Taipei Has he promised us the latest GPUs??? $WYFI to become market leader imminently
5
5
3
31
2,743
WhiteFiber should be up more.
1
2
10
515
It was a busy and exciting week at @AIInfraSummit. Great content, meetings, and introductions. Thanks to the organizers for having us and giving our engineers a forum to share their expertise. Also, big thanks to @NVIDIA for hosting us on campus while we were in town. 🚀 $WYFI
6
11
2
86
6,223
During yesterday AI Infra Summit in Santa Clara CA, OpenAI confirmed MORE compute is needed for safer AI. $WYFI about to explode, and for good reason. Follow WhiteFiber’s journey with “Retrofit” Colocation, WhiteFibers own Cloud, asset light Cloud, Managed Services and the “BIG” one Redwood Continuum. Can’t wait for the Redwood Continuum launch next Wednesday @WhiteFiber_
OpenAI's Sachin Katti says the AI safety debate is an argument for more compute, not less, because alignment itself has to be trained "If anything, I think it actually double-underscores the need for more compute." "The way we will achieve safety and alignment for these very capable models is by actually spending more compute on building the safety models, the alignment models that will make sure that the frontier models are actually safe and aligned." "So the perhaps counterintuitive conclusion from that is we'll actually be spending more compute." "We'll obviously continue to train and infer these frontier models, but in addition we will be investing in more compute to make sure that these models are safe and aligned." "So it's perhaps a counterintuitive observation, but if anything, we believe that we will need more compute to make sure that future models are more safe and more aligned." _________ More takeaways from Sachin Katti's conversation with NVIDIA: firesidealpha.substack.com/p…
1
4
1
22
1,535
Mark Zuckerberg: “Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us.”
Last month I wrote about how we can build a positive and safe future for everyone: meta.com/thefutureisforevery… Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens. The reality is: - People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned. There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind. - Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well. Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built. - Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators. - Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well. I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.
384
1,153
80
17,065
1,187,793