![](\"https:\/\/fortune.com\/img-assets\/wp-content\/uploads\/2026\/06\/55325145115_d284ae4708_o-e1781031590778.jpg?w=2048\")
<\/p>\n<\/p>\n
After spending years racing to build ever-larger AI models, researchers and infrastructure providers are increasingly focused on a new problem: how to make those systems affordable enough to deploy at scale.\u00a0<\/p>\n
Sara Hooker, cofounder and CEO of startup AI lab Adaption, told the audience at Fortune Brainstorm Tech on Tuesday that most of today\u2019s AI is what she called \u201cmonolithic\u201d\u2014or stuck in time. That is, once a model is trained, the model\u2019s knowledge and capabilities are essentially fixed. If something changes in the world, or if the model learns something useful from users, that knowledge doesn\u2019t automatically become part of the model.<\/p>\n
\u201cYou need models that can evolve,\u201d she explained, \u201cotherwise you end up with massive inefficiencies.\u201d\u00a0<\/p>\n
Still, for now, scale does matter\u2014and the biggest models are not going away anytime soon, said Rodrigo Liang, CEO of AI chip company SambaNova, though there will be \u201cplenty of room for more efficient models to come in.\u201d For the time being, he explained, customers are left to struggle with the cost of scaling models; with energy-hungry infrastructure; and with finding enough AI chops.\u00a0<\/p>\n
But Hooker focused on what\u2019s next, saying that we\u2019re at an \u201cinflection point with massive urgency to change that curve\u201d or model size. Most people, she explained, intuitively understand that you shouldn\u2019t just apply the same model to all problems. \u201cProbably 90% of problems are very easy\u2014many things that you do in bulk processing, for example, you shouldn\u2019t be throwing a massive model at.\u201d\u00a0<\/p>\n
She argued that future AI systems will need to adapt continuously to new information and rapidly change their behavior, rather than relying on repeated calls to a fixed model\u2014a dynamic she said is contributing to the soaring API bills many companies are now experiencing. Today\u2019s enterprises are deploying agents at scale, but those agents often aren\u2019t learning from their mistakes, so companies are paying repeatedly\u2014in compute, API calls, and infrastructure costs\u2014for the same errors.\u00a0<\/p>\n
While model developers like Hooker are focused on building more capable and efficient AI systems, Liang said the industry\u2019s immediate challenge is running today\u2019s massive models efficiently enough to make real-world deployments economically viable. He argued that trillion-parameter models remain too expensive and power-hungry, and said SambaNova\u2019s strategy is focused on delivering faster inference with lower power consumption through hardware specifically designed for large-model workloads.<\/p>\n
\u201cWe\u2019re getting two to 3x better than the [Nvidia] Blackwells [GPUs] on the exact same models, and so we think that at scale that\u2019s the way to at least bring the cost down,\u201d he said.<\/p>\n
More from Fortune\u2019s 25th Brainstorm Tech:<\/strong><\/em><\/p>\n Anthropic\u2019s Boris Cherny, creator of Claude Code, says there are days he manages tens of thousands of AI agents at once<\/em><\/p>\n \u2018Not an Allbirds Moment\u2019: Xbox\u2019s new CEO says she is grounding the console in gaming roots not AI<\/em><\/p>\n Anduril CEO Brian Schimpf says economic warfare is the \u2018new normal\u2019 for military conflicts\u2014and the U.S. needs to get serious<\/em><\/p>\n<\/div>\n #Adaption #CEO #Sara #Hooker #models #learn #continuously #reduce #soaring #costs<\/p>\n","protected":false},"excerpt":{"rendered":" After spending years racing to build ever-larger AI models, researchers and infrastructure providers are increasingly focused on a new problem: how to make those systems affordable enough to deploy at… <\/p>\n","protected":false},"author":1,"featured_media":6483,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[8913,3849,369,8916,449,8915,96,2453,1297,8914,8912,2361],"class_list":["post-6482","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-finance-news","tag-adaption","tag-brainstorm-tech","tag-ceo","tag-continuously","tag-costs","tag-hooker","tag-learn","tag-models","tag-reduce","tag-sara","tag-semiconductor-industry","tag-soaring"],"_links":{"self":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts\/6482","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6482"}],"version-history":[{"count":0,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts\/6482\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/media\/6483"}],"wp:attachment":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6482"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6482"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6482"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}