![\"Nvidia's](\"https:\/\/www.europesays.com\/ie\/wp-content\/uploads\/2025\/12\/1764829697_273_0x0.jpg\")
<\/p>\n<\/p>\n
Nvidia’s headquarters in Santa Clara, Ca.<\/p>\n
Nvidia<\/p>\n
OK, I haven\u2019t done this in a while; no excuse other than laziness. But here are ten concrete, defensible predictions for AI in 2026, with a bias toward things that materially matter for infra, enterprises, and policy.<\/p>\n
1. Agentic AI moves from demos to staffed \u201cdigital teams\u201d<\/p>\n
We\u2019ve all heard the cry, \u201cThe Agents are coming! The Agents are coming!\u201d, and this will be the largest theme in AI next year. By late 2026, many large enterprises will have at least one production agentic workflow (not just a POC) handling end\u2011to\u2011end tasks in support, finance, or operations.<\/p>\n
Analyst estimates put the autonomous AI agent market around the mid\u2011single digit billions in 2026, growing several\u2011fold by 2030, so this will be visible in expense and capital budgets.<\/p>\n
2. Inference dominates AI capex and reshapes hardware mix<\/p>\n
Ok, a no-brainer. Trying to up my average, here. Many estimates predict that by 2026, roughly two\u2011thirds of AI compute cycles will be spent on inference, not training, and most of that will still sit in data centers and enterprise servers rather than at the edge. Consequently, specialized inference in\u2011house silicon will grow much faster than GPUs in shipment growth, even if GPUs remain larger in absolute dollars. <\/p>\n
Nvidia is not about to be slain by ASICs, however. The Rubin CPX<\/a> demonstrates that Nvidia is ready for the high-memory future of agentic AI inference. Google TPUv7<\/a> will also be a serious contender, as will the Qualcomm AI200\/250<\/a> whose massive memory capacity and lower costs will be compelling. AMD will have to await Helios<\/a> in order to play, but other than price, their differentiation remains unclear. <\/p>\n 3. Domain\u2011specific and smaller models beat general AI LLMs in the enterprise<\/p>\n This one is also a no-brainer. Almost everyone agrees with this. <\/p>\n More than half of serious enterprise deployments will standardize on domain\u2011specific or fine\u2011tuned models (finance, healthcare, legal, telco) rather than a single general LLM, driven by accuracy, governance, and especially ROI.<\/p>\n 4. CSP and hyperscaler ASICs start to bite into Nvidia, especially Google.<\/p>\n The cloud\u2011native accelerators from hyperscalers (TPU, Trainium\/Inferentia, MTIA, etc.) plus Chinese incumbents will post materially (3X?) higher growth rates than merchant GPUs in 2026, driven by cost\u2011per\u2011token and ecosystem control. We certainly seem to be heading towards a structurally more competitive accelerator landscape. But I suspect Nvidia will remain supply-constrained, not demand, through at least 2026. And Rubin Ultra could extend this enviable state at least through 2027. <\/p>\n 5. Data, especially synthetic data, becomes as important as model architecture in AI<\/p>\n With credible forecasts that high\u2011quality public web data for training will be largely tapped out around 2026, leading labs and enterprises will invest aggressively in synthetic data pipelines and private data curation. 6. We will see the first AI \u201cWorld Models\u201d appear<\/p>\n \u201cWorld models\u201d here means models that build a structured internal representation of 3D\/physical environments and dynamics, often for simulation, robotics, or rich video, trained on multi-modal data, not the internet. World Labs, Google, and Meta will benefit. Qualcomm AI200\/250 could be a real surprise here as well. <\/p>\n 7. Identity theft and AI deep fakes will scare the ^&*! out of a lot of us.<\/p>\n High\u2011quality, real\u2011time deepfakes and AI\u2011assisted fraud (\u201cCEO doppelg\u00e4nger,\u201d voice clones, synthetic documents) will force enterprises to treat identity and content authenticity as first\u2011class security domains. This will certainly become a major issue in the upcoming mid-term elections. <\/p>\n 8. Consequently, AI regulation shifts from principle to enforcement and liability<\/p>\n In 2026, the regulatory conversation will move from broad frameworks to concrete enforcement: incident reporting, fines, and mandatory controls for high\u2011risk AI uses in finance, healthcare, employment, and critical infra. <\/p>\n Procurement and compliance teams will start to demand detailed model documentation, data lineage, and risk controls, affecting which vendors can sell into large accounts.<\/p>\n 9. AI hardware roadmaps double down on memory and interconnect, not just FLOPs<\/p>\n Next\u2011gen accelerators (GPUs, NPUs, CSP ASICs) launching in 2026 will emphasize HBM4, advanced packaging, and proprietary interconnect fabrics (NVLink\u2011class, vendor\u2011specific protocols), as the locus of competition moves from raw TOPS to system\u2011level throughput and scale\u2011out efficiency. <\/p>\n 10. AI productivity gains show up unevenly, widening the \u201cAI gap\u201d between firms<\/p>\n Most organizations will report using generative AI somewhere, but only a small minority will have it fully scaled across workflows with measurable ROI, creating a widening performance gap between AI\u2011mature and AI\u2011experimental firms. <\/p>\n Any Black Swans? <\/p>\n Of course there will be surprises. General AI will not be one of them. But that is what keeps this market so interesting! <\/p>\n Note that Nvidia and Qualcomm are clients of Cambrian-AI. <\/p>\n","protected":false},"excerpt":{"rendered":"Nvidia’s headquarters in Santa Clara, Ca. Nvidia OK, I haven\u2019t done this in a while; no excuse other…\n","protected":false},"author":2,"featured_media":214554,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[261],"tags":[291,7101,289,290,1797,116787,18,823,19,17,292,55409,82],"class_list":{"0":"post-214553","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-amd","10":"tag-artificial-intelligence","11":"tag-artificialintelligence","12":"tag-aws","13":"tag-broadcomm","14":"tag-eire","15":"tag-google","16":"tag-ie","17":"tag-ireland","18":"tag-nvidia","19":"tag-predictions","20":"tag-technology"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@ie\/115659886373434951","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/214553","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/comments?post=214553"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/214553\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media\/214554"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media?parent=214553"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/categories?post=214553"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/tags?post=214553"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
Benchmarks and R&D narratives will increasingly differentiate on data generation, filtering, and feedback loops rather than just model size.<\/p>\n