Hi All
Welcome to our brief overview of portfolio news from the past few days.
AMD and OpenAI Announce Landmark AI Partnership
AMD and OpenAI have announced a multi-year strategic partnership to power OpenAI’s next-generation AI infrastructure, sending AMD stock 35% higher pre-market. Under the agreement, OpenAI will deploy 6 gigawatts of AMD GPUs across multiple generations of the AMD Instinct family. The initial 1 gigawatt deployment, featuring the upcoming AMD Instinct MI450 Series GPUs, is set to begin in the second half of 2026.
For AMD, the financial implications are massive. The company expects the partnership to deliver tens of billions of dollars in revenue to its data center business.
Lisa Su :
“this partnership will enable additional revenue from existing and new customers deploying at scale and has the potential to generate well over $100 billion in revenue over the next few years”
“MI450 is a significant, significant step up, very competitive from a technology standpoint. This is a huge milestone for us. We’ve said that we believe that our data center AI revenue could be tens of billions of dollars going forward. You guys have often asked me when. I think we have a clear line of sight to this being achieved in 2027.”
Beyond a simple hardware purchase, the deal is structured as a deeply integrated partnership. To align strategic interests, AMD has issued OpenAI a warrant for up to 160 million shares of AMD (approximately 10% of shares capital). While this may initially seem like a high cost to acquire a customer, the vesting of these shares is tied to specific deployment milestones and AMD achieving certain share-price targets, creating win-win incentive structure.
On the negative side, this deal also highlights the vicious capital cycle powering the AI boom, echoing Nvidia’s previously announced 10-gigawatt and $100 billion investment plan with OpenAI.
AMD and IBM Collaborate on AI Infrastructure
AMD and IBM announced a collaboration to provide AI infrastructure to Zyphra, an open-source AI research company. Under the multi-year agreement, IBM will deliver a large cluster of AMD Instinct MI300X GPUs on IBM Cloud for Zyphra to use for training its frontier multimodal foundation models.
This is the first large-scale, dedicated training cluster on IBM Cloud to use AMD’s full-stack platform, including its advanced compute and networking hardware like AMD Pensando Pollara 400 AI NICs and AMD Pensando Ortano DPUs.
The deal marks a significant win for AMD, as it showcases the company’s ability to deliver a full-stack solution for large-scale AI training, not just inference. By proving its capability to handle end-to-end training for frontier models, AMD is better positioned to compete for billion-dollar datacenter contracts and capture market share from Nvidia.
OpenAI’s Sora 2 Sets a New Bar for Generative Video
OpenAI has released Sora 2, its latest and most powerful video generation model. The update marks a significant leap in quality, featuring more physically accurate and realistic video, along with new capabilities like synchronized dialogue and sound effects.
Alongside the model, OpenAI launched a new, invite-only social iOS app called “Sora,” which allows users to create and share short-form videos. This appears to be another “ChatGPT moment” for OpenAI, as the app’s launch was a massive success, immediately shooting to the No. 1 spot on the U.S. free App Store and displacing both Google’s Gemini and OpenAI’s own ChatGPT app.
This release intensifies the competitive pressure on Meta and Google, both of whom are developing their own generative video tools. The launch of Sora 2 just one week after Meta’s “Vibes” sets a new, very high benchmark for the industry, leaving Meta playing catch-up once again.
Currently, Sora aims to monetize the app by requiring users to pay for extra video generations, but it wouldn’t be surprising to see an advertising model introduced if the app gains massive adoption. What happens next remains to be seen, but one thing seems certain: compute demand will continue to increase exponentially while competition for consumers’ time in the social media space intensifies.
Greggs Q3 Trading Update
Greggs reported total sales growth of 6.1% for the third quarter, with company-managed like-for-like (LFL) sales up 1.5%. The company noted that trading improved in August and September after performance in July was held back by unusually high temperatures.
This LFL growth reflects a deceleration from the 2.6% first half of the year. Our bull thesis on the company is predicated on LFL sales accelerating above inflation, which is key to sustaining profitability margins. Adding to concerns about growth saturation, the company lowered its net new shop opening guidance for 2025 to around 120, citing the “timing of opportunities”. However, it’s difficult to judge based on one year’s performance, especially considering the UK’s macroeconomic weakness and cost inflation.
Despite these challenges, the stock price closed 6% higher after the results, indicating that much of the pessimism may have already been baked in.
On a positive note, management highlighted that the outlook for cost inflation has marginally improved. Key supply chain investments in Derby (planned to open in 2026) and Kettering (operational in 2027) are progressing as planned.
That's a wrap. See you soon.