--- title: "AMD's inference software caught up with Nvidia's, and its stock still fell 5.57%" description: "Two independent testers found AMD's inference software closing on Nvidia's; in the same week more than $1 trillion came off the most valuable AI and chip stocks on financing fears. The two moves price different things, and only one of them is about AMD's product." publisher: "The Inference" section: "Compute" published: 2026-10-15T17:26:38.712Z modified: 2026-10-15T17:26:38.712Z canonical: https://theinference.org/article/amd-s-inference-software-caught-up-with-nvidia-s-and-its-stock-still-fell-5-57 language: en keywords: "AMD, Nvidia, Inference, Benchmarks, AI Chips" --- # AMD's inference software caught up with Nvidia's, and its stock still fell 5.57% > Two independent testers found AMD's inference software closing on Nvidia's; in the same week more than $1 trillion came off the most valuable AI and chip stocks on financing fears. The two moves price different things, and only one of them is about AMD's product. Two independent testing shops re-scored [AMD](https://theinference.org/markets/companies/amd)'s Instinct MI355X accelerator this week, and both found the software gap with [Nvidia](https://theinference.org/markets/companies/nvidia) closing [1][2]. The stock fell anyway: the most valuable AI and semiconductor names [shed more than a trillion dollars during the week on borrowing-cost fears](https://www.cnbc.com/2026/10/15/sk-hynix-samsung-softbank-stocks-asia-tech-rout-ai-funding-worries.html), and AMD fell harder than Nvidia [3]. It was down 5.57% in early trading on October 15 [4]. The benchmark results measure what the chip can do, while the sell-off prices the cost of financing the data centers that buy such chips [3]. Only one of those moves is about AMD's product. ## The software link finally moved Artificial Analysis added the MI355X to AA-AgentPerf, its benchmark for agentic workloads where AI takes multi-step action instead of answering one prompt, and [reported on October 14 that the chip's results had improved substantially since the benchmark's June launch](https://cryptobriefing.com/amd-mi355x-aa-agentperf-deepseek-results/) on hardware that never changed: every gain came from software [1]. The split result survives. Nvidia still leads on efficiency on the lighter DeepSeek model, but on the larger DeepSeek V4 Pro the MI355X beats the B200 outright, an edge the testers credit to its 288 GB of memory per GPU [1]. *Chart: **On the lighter model, Nvidia still leads on tokens per megawatt** Bar chart comparing inference efficiency on DeepSeek V4.1 Flash: AMD MI355X at about 128 million tokens per second per megawatt, Nvidia B200 at about 193 million, Nvidia B300 at about 220 million* *Output throughput per megawatt at 100 tokens per second per user, DeepSeek V4.1 Flash, single 8-GPU node on ROCm nightly builds; approximate values as reported by Artificial Analysis, October 14, 2026. [1]* ## By September, AMD was the cheaper chip to run SemiAnalysis, whose own host admits the firm was long anti-AMD, watched AMD's ATOM inference engine catch and briefly overtake some Nvidia configurations on the GLM 5.3 coding model as engineers optimized it over weeks [2]. [Its September work put the MI355X about 9% cheaper to run than Nvidia's B200](https://www.benzinga.com/markets/prediction-markets/26/10/62355815/amd-nvidia-ai-inference-software-gains) when responses must begin within two seconds [2]. On MLPerf, an industry benchmark suite, AMD reported squeezing up to 38% more work from the same chips [2]. Cloud operators paying the inference bill finally have a benchmarked second source. The gap is being closed by AMD's own engineers, who now ship updates to ROCm, the company's software platform for its GPUs, roughly every six weeks, with AI agents helping write the code [2]. The company's [Helios racks entered production in September](https://www.techtimes.com/articles/329005/20261014/amd-helion-ocp-blueprint-adds-model-router-operators-scale-cpus-without-buying-more-gpus.htm) and are installed at [OpenAI](https://theinference.org/markets/companies/openai), [Meta](https://theinference.org/markets/companies/meta), [Microsoft](https://theinference.org/markets/companies/microsoft) and [Oracle](https://theinference.org/markets/companies/oracle) [5]. ## Routing decides who serves the agents Agentic work multiplies model calls, so cost per token decides whose silicon serves them. AMD's answer is Helion, [a rack blueprint it published at the Open Compute Project summit on October 12](https://newsroom.amd.com/) and handed to the open-hardware community instead of selling it [6]. The blueprint embeds a model router in the rack: routine agent calls go to a cheap open-source model hosted locally, hard ones go to a frontier system, and AMD claims a 43% cut in token costs from internal testing that no one has verified at rack level [5]. [The Taalas purchase, confirmed the same week](https://www.diesignal.com/articles/amd-buys-taalas-to-add-model-specific-silicon-to-instinct-roadmap-0a9ab45c), points the same way: split inference across specialized chips, each doing the part it does best. Taalas is a startup that bakes one model's weights into chip circuitry so the model can be served cheaply, mirroring Nvidia's $20 billion purchase of the inference-chip maker [Groq](https://theinference.org/markets/companies/groq) in December 2025 [7]. ## The limiting link is now the balance sheet AMD's software, its weakest argument for two years, moved this week. The limit now is money. Investors are asking whether higher borrowing costs will slow the capital spending that buys these racks, a worry that applies to every chip vendor's customers at once [3]. As Dan Ives, tech analyst at Yorkville Ives, put it, rates are "a financing question, not a demand question" [3]. AMD shareholders carry the sharper version of that risk. They pay 22.6 times sales for the stock, a price that assumes the operating margin keeps climbing from the 17.3% of the June quarter [4]. [The next report, due around November 3](https://finance.yahoo.com/markets/stocks/articles/risk-hiding-amd-stocks-big-133132787.html), lands in the middle of the scare [4]. *Chart: **AMD's operating margin has climbed from near zero to 17.3% of sales** Line chart of AMD quarterly operating margin from Q3 2023 to Q2 2026, rising from 3.9 percent to 17.3 percent with a dip to minus 1.7 percent in Q2 2025* *Operating margin as a share of revenue, fiscal quarters to June 2026. Sources: AMD SEC filings via company_series; the 2025 loss quarter carried roughly $800 million of MI308 export-related charges. [8][4]* AMD's benchmark numbers improved this week without a new chip. What the share price rests on now is the margin, and the November report is the first reading of it. ## Takeaway AMD's inference software closed on Nvidia's again this week on unchanged hardware, but the stock now trades on financing costs and a 22.6-times-sales valuation, and the November 3 report is the first test of the margin story. ## Sources 1. [Crypto Briefing, "AMD MI355X joins AA-AgentPerf with DeepSeek V4.1 Flash results", 14 October 2026](https://cryptobriefing.com/amd-mi355x-aa-agentperf-deepseek-results/) 2. [Benzinga, "AMD Challenges Nvidia in AI Inference, SemiAnalysis Finds", October 2026](https://www.benzinga.com/markets/prediction-markets/26/10/62355815/amd-nvidia-ai-inference-software-gains) 3. [CNBC, "Nvidia, AMD, Micron fall as AI stock sell-off deepens", 15 October 2026](https://www.cnbc.com/2026/10/15/sk-hynix-samsung-softbank-stocks-asia-tech-rout-ai-funding-worries.html) 4. [Trefis via Yahoo Finance, "Is There A Risk Hiding In AMD Stock's Big Run?", 15 October 2026](https://finance.yahoo.com/markets/stocks/articles/risk-hiding-amd-stocks-big-133132787.html) 5. [TechTimes, "AMD Helion OCP Blueprint Adds Model Router: Operators Scale CPUs Without Buying More GPUs", 14 October 2026](https://www.techtimes.com/articles/329005/20261014/amd-helion-ocp-blueprint-adds-model-router-operators-scale-cpus-without-buying-more-gpus.htm) 6. [AMD Newsroom, "Introducing AMD Helion, Agentic AI Compute at Rack Scale", 12 October 2026](https://newsroom.amd.com/) 7. [Die Signal, "AMD Buys Taalas to Add Model-Specific Silicon to Instinct Roadmap", October 2026](https://www.diesignal.com/articles/amd-buys-taalas-to-add-model-specific-silicon-to-instinct-roadmap-0a9ab45c) 8. [AMD, quarterly figures from its SEC filings (retrieved via company_series, 15 October 2026)](https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000002488) --- AMD's inference software caught up with Nvidia's, and its stock still fell 5.57% — The Inference. Canonical: https://theinference.org/article/amd-s-inference-software-caught-up-with-nvidia-s-and-its-stock-still-fell-5-57