Senior Vice President, Global AI Markets at AMD
Orange County, CA
Joined January 2013
- Tweets5.1K
- Following1.1K
- Followers1.4K
- Likes1.2K
Keith Strier retweeted
$AMD OpenAI VP/Head of Hardware Richard Ho said in an interview that OpenAI is deploying Jalapeño with AMD EPYC Turin hosts, each with 1.5 TB of DRAM. Ho called the choice “pragmatic” and cited Turin’s maturity versus Nvidia Vera.
This news alone should add $300B+ to AMD's market cap. As custom AI silicon proliferates, the CPU controlling memory, networking and orchestration remains valuable asset. AMD now can monetize the shift away from GPUs into this CPU super cycle.
Keith Strier retweeted
This is the most powerful workstation in the world.
Designed and engineered for a completely new era of computing.
A machine capable of running AI models with more than a trillion parameters right on your desk.
We call it the Threadripper Halo Station.
• 96 cores of Threadripper PRO
• 2 MI350P Instinct accelerators with a path to 4
• Up to 2 TB of system memory
• Up to 576 GB of HBM3e
• Liquid cooled so all that power stays super quiet
This is about as close as you can get to a personal supercomputer.
ALT AMD SVP & GM Jack Huynh unveiling the most powerful workstation in the world at his IFA opening keynote in Berlin, the Threadripper Halo Station.
Keith Strier retweeted
¿Cómo puede la #IA transformar a México? 🤖🇲🇽
@kbsdigital, vicepresidente senior de Mercados Globales de Inteligencia Artificial de @AMD, comparte su visión sobre el papel de la IA, el desarrollo del talento y las oportunidades que se abren para el futuro de nuestro país.
Keith Strier retweeted
"Each generation has seized its opportunity to expand the frontiers of knowledge & lay down the paving stones of progress. Now it is our time to build the scientific enterprise that will carry America forward and bring prosperity to Americans through the 21st century & beyond" 🇺🇸
Keith Strier retweeted
We want AMD delightful to use, both for software and hardware. Great work @jackhuynh and team on the Halo box
The Ryzen Halo by @AMD is a work of art, it's quiet and cool, has a subtle but slick Need for Speed Underground aesthetic, blazing fast, most AI software worked OOB. Between this & the @FrameworkPuter desktop, AMD is shipping the most beautiful electronic devices on the market.
Keith Strier retweeted
Great spending time with Rick Mershad, CEO of @microcenter, to talk about an exciting milestone for builders.
Ryzen AI Halo systems are now shipping from Micro Center.
💪 @AMD is the #1 desktop CPU at Micro Center with Ryzen X3D.
🎮 Radeon powers approximately 1 in every 3 desktop gaming GPUs sold at Micro Center.
Now we are extending that same momentum into a new era.
Ryzen AI Halo gives AI developers and builders the ability to run larger models locally, iterate faster, and unlock powerful new AI experiences while keeping more of their work on device for greater privacy and security, without cloud limits.
The Agentic PC is here 🚀
Keith Strier retweeted
Honored to join @POTUS, @InvestAmerica24 and leaders from across industry and government at the White House today. As America celebrates 250 years, @AMD is proud to support @TrumpAccounts and help build a stronger foundation for our employees' families and the next generation of American innovators.
Keith Strier retweeted
$AMD powers 4 of the world’s top 10 supercomputers, more than any other chip company on the latest TOP500 list
Keith Strier retweeted
$AMD & Oriole Networks work to beat $NVDA GPU🚀
The Oriole Networks and AMD £50 million (~$68M) UK ARIA Scaling Inference Lab delivers significant strategic benefits to AMD by deploying its Instinct GPUs and EPYC CPUs in the world’s first large-scale pure photonic AI network. This high-profile project showcases AMD hardware in a revolutionary networking environment that dramatically improves performance, achieving an 81% reduction in network power consumption, slashing GPU idle time from ~60% to under 1%, and delivering up to 10x higher inference throughput and interactivity.
By solving critical communication bottlenecks that plague traditional GPU clusters, AMD’s accelerators appear far more powerful and efficient, strengthening its competitive position against NVIDIA’s dominant networking solutions. The partnership provides real-world validation for AMD in next-generation, xPU-agnostic photonic fabrics, lowers total cost of ownership for customers through major energy and cooling savings, and creates a compelling sales pipeline as Oriole scales deployments in 2027. Overall, AMD gains market credibility as an innovator in sustainable, high-performance AI infrastructure, boosts demand for its chips in frontier-scale inference, and positions itself at the forefront of energy-efficient computing.
Source : quantumzeitgeist.com/oriole-…
BREAKING $AMD Agentic AI Rack At SCALE is out! 🚀
AKA CPUs Dense Rack for Fleets of Agents!🆕🆕🆕
I been talking about this for years & it is HAPPENING
Not Financial Advice! DYOR!
Today, @AMD published a detailed technical blog emphasizing that the future of agentic AI autonomous, multi-step AI systems requiring heavy orchestration, databases, caching, APIs, and control planes demands massive CPU-dense rack-scale infrastructure, not just GPUs. The catalyst prominently positions their upcoming 6th Gen EPYC "Venice" processors as the key enabler for next-generation dense racks, delivering leadership throughput under real-world power, cooling, and density constraints.
AMD's EPYC-powered "dense rack" (primarily the Helios platform with Venice CPUs) is on track for volume availability and deployments in the second half of 2026 (H2 2026).
Timeline:
Venice production ramp May 2026
Helios Previews/showcases Jan-June 2026
Customers testing Q2-Q3
Vol deployments From H2 2026
Accelerated Ramp Q4 2026 and onward
UBS Note is completely false and misleading!
Key highlights:
~EPYC Venice (Zen 6 architecture, up to 256 cores / 512 threads per socket) is projected to deliver exceptional rack-level performance. In AMD’s modeled 100 kW rack comparisons, Venice-powered systems are expected to achieve ~3.30x the throughput of NVIDIA’s Vera (88-core Olympus) baseline across a broad mix of agentic-supporting workloads.
~This builds on current-generation 5th Gen EPYC "Turin" (up to 192 cores), which already delivers ~2.37x rack throughput vs. Vera and ~1.6x vs. Intel’s Xeon 6980P (128 cores).
~ Liquid-cooled Turin deployments already support >27,000 CPU cores per rack today. Venice is architected to push this beyond 36,000 cores in the same rack class, dramatically increasing concurrent agent capacity and overall infrastructure efficiency.
1. The Helios AI Rack Platform
Venice CPUs are a cornerstone of AMD’s Helios reference design, the company’s flagship open, standards-based rack-scale AI platform. Helios integrates:
~EPYC Venice CPUs for orchestration and CPU-bound services.
~Instinct MI400-series GPUs ( MI455X) for AI acceleration.
~Pensando Vulcano AI networking for high-bandwidth, low-latency scale-out.
Each compute node in Helios typically pairs one Venice CPU with four MI455X GPUs, enabling full liquid cooling and extreme density. The platform is on track for multi-gigawatt deployments in the second half of 2026, targeting hyperscalers and large enterprises running production agentic AI
2. Key Performance Advantages for Agentic AI
AMD’s analysis normalizes everything to 100 kW racks using dual-processor (2P) servers, focusing on real deployable capacity rather than isolated chip benchmarks. Workloads modeled include:
~SPEC CPU 2017 Integer Rate
~Server-side Java
~NGINX web serving
~Redis / Memcached
~MySQL OLTP (TPROC-C
Venice is also projected to maintain strong single-threaded/per-core performance (critical for latency-sensitive database and orchestration tasks), with up to +27% per-core advantage vs. competitors at comparable configurations. Additional benefits include up to 1.6 TB/s memory bandwidth per socket and advanced I/O.
Conclusion:
AMD stresses deployability today on standard x86 platforms (no proprietary architectures required), full software compatibility, and open standards. This positions Venice + Helios as a practical, high-density alternative to competing solutions while underscoring that agentic AI shifts the balance toward CPU-rich racks alongside GPUs.
In her recent public remarks and earnings commentary, AMD Chair and CEO Dr. Lisa Su has been clear and bullish: the rapid rise of inference-heavy, agentic AI workloads represents a major inflection point, one that is poised to trigger a steep J-curve acceleration in demand for high-performance CPUs and balanced CPU+GPU infrastructure.
“We always believed that inference would actually be the driver of AI going forward, and we can now see that inference inflection point,” Su stated. She emphasized that as AI adoption scales from simple chatbots to autonomous agentic systems, which continuously orchestrate tasks, access data, reason over long horizons, and operate in real time the compute requirements shift dramatically. These workloads demand far more CPU resources for orchestration, data movement, caching, databases, and head-node control planes alongside GPU acceleration.
This shift is already reshaping the economics of AI infrastructure. The current CPU TAM forecast to more than $500 billion by 2030 where analysts are expecting AMD to be biggest winner. Dr. Su noted that customer demand across cloud and enterprise is running “stronger than expected,” with agentic AI and large-scale inference driving both larger GPU clusters and significantly higher CPU density per rack.
The result is a classic J-curve for AMD: modest but visible gains today from 5th Gen EPYC Turin dense racks, followed by a sharp upward trajectory as 6th Gen EPYC Venice and the Helios rack-scale platform (Venice CPUs + MI400-series GPUs) ramp in the second half of 2026. With Venice enabling >36,000 CPU cores per liquid-cooled rack and Helios delivering rack-level performance leadership, AMD is positioning itself at the center of the next phase of AI buildout, one where balanced, CPU-rich, open-standards infrastructure becomes the dominant architecture.
Dr. Su’s message is confident: the industry is still in the “early innings” of unlocking AI’s full potential, with global compute needing to grow ~100x in the coming years to reach yottascale. For AMD, the combination of leadership in EPYC server CPUs, a strong Instinct GPU roadmap, and open rack-scale solutions like Helios places the company squarely on the steep part of the J-curve.
Not Financial Advice! DYOR!
AMD Blog Source: amd.com/en/blogs/2026/agenti…
Keith Strier retweeted
BREAKING $AMD Agentic AI Rack At SCALE is out! 🚀
AKA CPUs Dense Rack for Fleets of Agents!🆕🆕🆕
I been talking about this for years & it is HAPPENING
Not Financial Advice! DYOR!
Today, @AMD published a detailed technical blog emphasizing that the future of agentic AI autonomous, multi-step AI systems requiring heavy orchestration, databases, caching, APIs, and control planes demands massive CPU-dense rack-scale infrastructure, not just GPUs. The catalyst prominently positions their upcoming 6th Gen EPYC "Venice" processors as the key enabler for next-generation dense racks, delivering leadership throughput under real-world power, cooling, and density constraints.
AMD's EPYC-powered "dense rack" (primarily the Helios platform with Venice CPUs) is on track for volume availability and deployments in the second half of 2026 (H2 2026).
Timeline:
Venice production ramp May 2026
Helios Previews/showcases Jan-June 2026
Customers testing Q2-Q3
Vol deployments From H2 2026
Accelerated Ramp Q4 2026 and onward
UBS Note is completely false and misleading!
Key highlights:
~EPYC Venice (Zen 6 architecture, up to 256 cores / 512 threads per socket) is projected to deliver exceptional rack-level performance. In AMD’s modeled 100 kW rack comparisons, Venice-powered systems are expected to achieve ~3.30x the throughput of NVIDIA’s Vera (88-core Olympus) baseline across a broad mix of agentic-supporting workloads.
~This builds on current-generation 5th Gen EPYC "Turin" (up to 192 cores), which already delivers ~2.37x rack throughput vs. Vera and ~1.6x vs. Intel’s Xeon 6980P (128 cores).
~ Liquid-cooled Turin deployments already support >27,000 CPU cores per rack today. Venice is architected to push this beyond 36,000 cores in the same rack class, dramatically increasing concurrent agent capacity and overall infrastructure efficiency.
1. The Helios AI Rack Platform
Venice CPUs are a cornerstone of AMD’s Helios reference design, the company’s flagship open, standards-based rack-scale AI platform. Helios integrates:
~EPYC Venice CPUs for orchestration and CPU-bound services.
~Instinct MI400-series GPUs ( MI455X) for AI acceleration.
~Pensando Vulcano AI networking for high-bandwidth, low-latency scale-out.
Each compute node in Helios typically pairs one Venice CPU with four MI455X GPUs, enabling full liquid cooling and extreme density. The platform is on track for multi-gigawatt deployments in the second half of 2026, targeting hyperscalers and large enterprises running production agentic AI
2. Key Performance Advantages for Agentic AI
AMD’s analysis normalizes everything to 100 kW racks using dual-processor (2P) servers, focusing on real deployable capacity rather than isolated chip benchmarks. Workloads modeled include:
~SPEC CPU 2017 Integer Rate
~Server-side Java
~NGINX web serving
~Redis / Memcached
~MySQL OLTP (TPROC-C
Venice is also projected to maintain strong single-threaded/per-core performance (critical for latency-sensitive database and orchestration tasks), with up to +27% per-core advantage vs. competitors at comparable configurations. Additional benefits include up to 1.6 TB/s memory bandwidth per socket and advanced I/O.
Conclusion:
AMD stresses deployability today on standard x86 platforms (no proprietary architectures required), full software compatibility, and open standards. This positions Venice + Helios as a practical, high-density alternative to competing solutions while underscoring that agentic AI shifts the balance toward CPU-rich racks alongside GPUs.
In her recent public remarks and earnings commentary, AMD Chair and CEO Dr. Lisa Su has been clear and bullish: the rapid rise of inference-heavy, agentic AI workloads represents a major inflection point, one that is poised to trigger a steep J-curve acceleration in demand for high-performance CPUs and balanced CPU+GPU infrastructure.
“We always believed that inference would actually be the driver of AI going forward, and we can now see that inference inflection point,” Su stated. She emphasized that as AI adoption scales from simple chatbots to autonomous agentic systems, which continuously orchestrate tasks, access data, reason over long horizons, and operate in real time the compute requirements shift dramatically. These workloads demand far more CPU resources for orchestration, data movement, caching, databases, and head-node control planes alongside GPU acceleration.
This shift is already reshaping the economics of AI infrastructure. The current CPU TAM forecast to more than $500 billion by 2030 where analysts are expecting AMD to be biggest winner. Dr. Su noted that customer demand across cloud and enterprise is running “stronger than expected,” with agentic AI and large-scale inference driving both larger GPU clusters and significantly higher CPU density per rack.
The result is a classic J-curve for AMD: modest but visible gains today from 5th Gen EPYC Turin dense racks, followed by a sharp upward trajectory as 6th Gen EPYC Venice and the Helios rack-scale platform (Venice CPUs + MI400-series GPUs) ramp in the second half of 2026. With Venice enabling >36,000 CPU cores per liquid-cooled rack and Helios delivering rack-level performance leadership, AMD is positioning itself at the center of the next phase of AI buildout, one where balanced, CPU-rich, open-standards infrastructure becomes the dominant architecture.
Dr. Su’s message is confident: the industry is still in the “early innings” of unlocking AI’s full potential, with global compute needing to grow ~100x in the coming years to reach yottascale. For AMD, the combination of leadership in EPYC server CPUs, a strong Instinct GPU roadmap, and open rack-scale solutions like Helios places the company squarely on the steep part of the J-curve.
Not Financial Advice! DYOR!
AMD Blog Source: amd.com/en/blogs/2026/agenti…
BREAKING $AMD UBS Bullshit Note 🚨🚨🚨
I refused to believe UBS Does not have access to BofA 2026 GLobal Tech Conference Call, Even myself, a Nobody got it on June 3rd 2026. This note from UBS is full of shit and lying bullshit. I been dealing with fake news like this for years, so get used to it.
Here is key points from UBS's post-Computex note (June 8-9 2026), after the BofA conference.
Nvidia's Timeline: Blackwell leads shipments in 2026–2027; industry checks align with their bullish outlook.
AMD's Timeline: The next-gen Helios platform will see motherboard shipments possibly in Q4 2026, but full data-center deployments are likely pushed to late 2027 due to extra testing and integration needs.
Ok there has been significant more delay news from the Media on $NVDA Rubin rack for the last 2-3 months. And this has nothing to do with @AMD!!!
Dr. Su already said Helios Rack is on track along with other AMD Executives.
On BofA Conference call, they said new customers are testing Helios Rack not Existing orders. OpenAI and $META are going to get their Helios Rack in H2 2026 and 4GW in 2027(Implying faster deployment).
NOT Late 2027 Bullshit.
Not Financial Advice! DYOR!
Source: tipranks.com/news/the-gap-re…
Keith Strier retweeted
We've made a breakthrough in self-evolving AI scientists moving from "search" to "principled discovery": Scientific discovery requires that the search space itself changes, and an AI scientist must perceive this shift without intervention. We built an AI that achieves this for the first time with the ability to discover the scientific vocabulary it reasons in. Evidence, tools, artifacts, verifiers, failures & claims become typed provenance. We show three distinct modalities: 1) retrieval, adding known objects; 2) search, exploring a fixed schema; and critically: 3) discovery, a verified regime transition.
We solve the open-endedness evaluation problem by lifting agentic workflows into a typed copresheaf and proving, via a Kan obstruction, that true discovery is not unbounded generation but a verifiable schema expansion: old evidence is transported by Left Kan extension, and genuine novelty is mathematically quantified by the pointwise residual beyond the transported image - separating discovery from mere search and making novelty objective and measurable rather than a subjective judgment or benchmark delta.
Our AI scientist is built in a way that does not pre-conceive the approach it chooses; instead, we endow the system with formal power to adapt, evolve, and reason from first principles. Case studies include:
1⃣Builder/Breaker model that discovers mode-conditioned compliance in proteins;
2⃣CategoryScienceClaw that finds anisotropic fiber-network stiffness rules.
Great work in collaboration with my graduate student @fwang108_ @MITdeptofBE
F.Y. Wang & M.J. Buehler, Self-Revising Discovery Systems for Science: A Categorical Framework for Agentic Artificial Intelligence, arXiv:2606.01444, 2026
Keith Strier retweeted
Энэ жилийн арга хэмжээний гол одод Цахиурын хөндийгөөс холбогдоно.
- AI-ийн тэсрэлтийн гол ялагчдын нэг @AMD-ийн SVP @kbsdigital
- Монголын мэдээллийн технологийн салбарын нүүр царай болсон @Google-ийн @battulga11
- Дэлхийн хамгийн том технологийн гигантуудын нэг @amazon-ийн Г.Мөнхбаяр
Keith Strier retweeted
🚀 Дэлхийн хагас дамжуулагчийн технологийн аварга @AMD компанийн Ахлах дэд ерөнхийлөгч @kbsdigital -ийн гал асаасан ярилцлага DevSummit 2026 арга хэмжээний тайзнаа!
Өнөөдөр дэлхийн технологийн салбарын хамгийн халуухан сэдэв бол гарцаагүй Agentic AI. Түүнийг ажиллуулах суурь хэрэгцээ болох GPU (График процессор)-ийн зах зээлд АНУ болон БНХАУ-ын өрсөлдөөн улам ширүүссээр байна.
⚡ Тэгвэл БНХАУ-тай хил залгаа, АНУ-тай "Гуравдагч хөрш"-ийн бодлоготой Монгол Улс энэ том давалгаанд хэрхэн тэнцвэрээ хадгалж, зөв нүүдэл хийх вэ?
Манай улсад ямар боломж, сорилт байгааг AMD компанийн Ахлах дэд ерөнхийлөгч (SVP) Keith Strier-тэй хийсэн онцгой ярилцлагаас олж мэдээрэй.
🔥 Зөвхөн DevSummit 2026 арга хэмжээний тайзнаас хүлээн авч үзэх боломжтой.
🗓️ Хэзээ: 5-р сарын 30, 31
📍 Хаана: Тусгаар Тогтнолын Ордон
🎟️ Тасалбар худалдан авах: devsummit.dev/tickets
Keith Strier retweeted
Q: How are job postings for software engineers rising rapidly despite AI agents automating coding?
A: Because there’s far more code to manage than ever before. We’re already seeing a 14x YoY increase in GitHub commits, and it’s accelerating.
AI has dramatically lowered the cost of writing code, so it’s now being used across far more businesses, applications, and use cases.
We’re at the beginning of a massive productivity boom driven by the proliferation of bespoke software throughout the entire economy.
Coding has been AI’s breakout use case this year. The fact that it’s increased demand for software engineers — rather than decreased it — should call into question the entire “AI will cause mass job loss” narrative.
Keith Strier retweeted
✨ Personal AI is the next computing platform.
AI is shifting from something you access to something you build with, locally, at the edge, and across systems.
We’re unlocking new possibilities for developers:
• @AMD Ryzen AI Halo, a local-first developer system, preorder starting in June, develop AI without limits on your desk
• Gorgon Halo with up to 192GB unified memory, supporting 300B+ parameter models locally, run massive models locally
We’re excited to partner with @ClementDelangue🤗, Co-founder and CEO of @huggingface, to advance open-source AI for Ryzen AI.
Our focus is seamless AI, from model to deployment.
Cloud, edge, device. One continuum.
Multi-agent systems, local inference at scale, open models as infrastructure.
This is the next computing era 🚀
Keith Strier retweeted
Built for the harshest environments.
Radiation-tolerant, space-grade AMD Virtex-5QV FPGAs were onboard the Artemis II Orion spacecraft, enabling functions critical to the success of the mission.
Details from @FierceSensors 🚀 fiercesensors.com/sensors/na…
