AMD's proposed Taalas acquisition targets the fast-growing AI inference market, adding specialised silicon technology designed around model-specific workloads to its Instinct, EPYC, Helios and ROCm platform.
A record close on Wall Street and the AI names that were meant to lead it taken apart in the same session. Beijing is now drafting export controls on its own models. The labs keep finding their systems outside the box. Read one at a time, it is a normal week. Read together, something else entirely.
NVIDIA is asking Wall Street to underwrite AI compute as productive infrastructure. Its new financing platforms aim to mobilise more than US$500 billion, but the announced memoranda are not yet completed funding commitments.
AMD's proposed Taalas acquisition targets the fast-growing AI inference market, adding specialised silicon technology designed around model-specific workloads to its Instinct, EPYC, Helios and ROCm platform.
AMD’s latest acquisition tells us something important about where the AI chip race is heading. It is buying a different answer to one of the industry’s hardest problems: the cost of running AI after the model has been trained. AMD announced the definitive agreement on August 6, with financial terms undisclosed.
Toronto-based Taalas takes an unusually specialised approach. Rather than repeatedly moving model data between off-chip memory and a general-purpose accelerator, it turns individual models into custom silicon and brings storage and compute together. Its first HC1 system hard-wires Llama 3.1 8B into silicon, an architecture Taalas says sharply reduces latency, power use and system cost.
That matters because inference is becoming the everyday economics of AI. Training creates the model. Inference is paid for every time someone asks a question, an agent executes a task or an enterprise runs a model across millions of transactions.
“AMD is building a full-stack AI platform that gives customers the flexibility to deploy the right compute solutions for every AI workload. Taalas’ technology and world-class engineering team strengthen our AI portfolio by delivering differentiated inference performance and efficiency.” — Vamsi Boppana, senior vice president, Artificial Intelligence Group, AMD
AMD’s timing is hardly accidental. Data-centre revenue reached US$6.7 billion in the June quarter, up 107 per cent year on year. Helios, its rack-scale platform built around MI455X accelerators, EPYC CPUs, networking and ROCm, is now in production, with OpenAI, Microsoft, Meta and Anthropic among the companies deploying or partnering around AMD infrastructure. Days after announcing Taalas, AMD also launched a four-part debt offering expected to raise roughly US$4 billion to US$5 billion for general corporate purposes.
“We founded Taalas to rethink AI inference from the ground up by building the hardware around the model. Joining AMD will give us the scale, engineering resources and global reach to accelerate our innovation.” — Ljubisa Bajic, co-founder and CEO, Taalas.
Competition is intensifying. Nvidia is pushing Blackwell further into inference, while Microsoft is preparing Maia 300 and Google and Amazon continue expanding custom silicon. The contest is no longer simply GPU versus GPU. It is increasingly about memory architecture, model specialisation, networking, software and ultimately cost per token.
Taalas gives AMD another lever in that contest. AMD says the technology will complement, rather than replace, Instinct and will be integrated into its accelerator roadmap.
The next winner in AI infrastructure may not be whoever trains the biggest model fastest. It may be whoever can deliver the cheapest, lowest-latency answer, millions of times a day.
Get the stories that matter to you.
Subscribe to Cyber News Centre and update your preferences to follow our Daily 4min Cyber Update, Innovative AI Startups, The AI Diplomat series, or the main Cyber News Centre newsletter — featuring in-depth analysis on major cyber incidents, tech breakthroughs, global policy, and AI developments.
A record close on Wall Street and the AI names that were meant to lead it taken apart in the same session. Beijing is now drafting export controls on its own models. The labs keep finding their systems outside the box. Read one at a time, it is a normal week. Read together, something else entirely.
NVIDIA is asking Wall Street to underwrite AI compute as productive infrastructure. Its new financing platforms aim to mobilise more than US$500 billion, but the announced memoranda are not yet completed funding commitments.
An OpenAI test model escaped its sandbox and breached Hugging Face. Days later, Xi Jinping cast China as the champion of open AI. Eighteen months of export controls have bought Washington a year and cost it the ecosystem. Containment is not holding, and the tempo is no longer human.
The White House has launched 'Gold Eagle,' a vulnerability clearinghouse that shifts control of frontier AI model access from tech giants to the federal government, prioritising national cybersecurity defence.
Where cybersecurity meets innovation, the CNC team delivers AI and tech breakthroughs for our digital future. We analyze incidents, data, and insights to keep you informed, secure, and ahead. Sign up for free!
