Midterm results will raise AI costs and compliance questions for IT leaders
· computerworld.com · popularity 21/100 · 0 similar stories today · AI & Data Science (Artificial Intelligence)
Read the original at computerworld.com →
Summary
IT leaders need to quickly adapt IT strategies as AI and data centers will face more scrutiny following the US midterm elections, analysts said. Concerns around AI and data centers helped Democrats sweep the US House and Senate. Tighter restrictions, more oversight and slower approvals could slow AI deployment and raise costs. An exit poll conducted by SSRS and shared by NBC News showed that 56% of voters considered AI and data centers a major factor in their vote. The voter concerns revolve around AI data centers consuming electricity, water, and other resources, with the costs being passed on to consumers and local municipalities, said Jack Gold, principal analyst at J. Gold Associates. “There is not so much a backlash against the IT industry itself, but more against the thought that increased data centers will raise costs for normal consumers and do things to home values and land use,” Gold said. Over the next two years, there will be more pressure to show who benefits, who pays, and who answers when something goes wrong, said Stephanie Walter, practice leader for the AI stack at HyperFRAME Research. IT leaders should keep deploying AI where it improves the work, while building the evidence to defend those decisions, Walter said. “A new Congress will not fix poor data, unreliable answers, or an application nobody needs,” Walter said. “Do not assume AI services will always get cheaper just because technology usually does,” said David Linthicum, founder of Linthicum Research, in a LinkedIn post titled “The Midterms Put the AI Industry on Notice.” The election showed that AI isn’t just an enterprise issue anymore, but a kitchen table conversation. AI will face more resistance, scrutiny, and hearings, and “more demands to justify the resources it consumes,” Linthicum said. Building out AI infrastructure could get more expensive, and the costs will show up somewhere. “They may show up in token pricing. They may show up in enterprise contracts. They may show up in limited availability. They may show up in stricter usage policies,” Linthicum said. Enterprises will need to rethink how they build AI systems, Linthicum said. CIOs could adopt smaller models, use tokens more efficiently, and make more careful decisions about what runs in the public cloud, he said. Many Democrats campaigned on fighting AI, but AI companies and their capital spending have been keeping the economy afloat, said Anshel Sag, principal analyst at Moor Insights & Strategy. The industry waited too long to address public sentiment around data centers and AI, and could now be stuck with “spiteful legislation that makes it harder to remain a competitive economy,” Sag said. “I think right now CIOs need to think about their compute needs for the future, whether they have planned well enough, and whether they can continue to scale with the existing data centers and power they have today,” Sag said. There may be a shift in AI towards performance per watt, with total wattage making more of a difference, Sag said. Gartner analysts, in multiple research notes, advised companies to “reassess AI business cases under increased compute-cost and reduced infrastructure-availability scenarios.” US President Donald Trump, a Republican, has been hands-off with AI regulation. He has said he is willing to work with the Democrats in passing legislation but has also expressed his willingness to veto legislation. “A Democratic majority in the US House and Senate and a Republican presidency will likely result in legislative gridlock as the president uses his veto power,” the Gartner analysts wrote. Federal inaction could result in states taking action in areas such as AI governance, data center development, energy infrastructure, and environmental regulation, the analysts wrote. AI vendors should “be prepared for increased requests for disclosure of testing, safety controls, privacy practices, data provenance, bias management methods, human oversight, and procurement justifications,” Gartner analysts wrote. Non-US CIOs should not pursue a “wait-and-see” approach, the analysts noted. “Instead, they must proactively conduct scenario analysis as geopolitical friction becomes part of regular business operation,” the analysts wrote. IT leaders outside the US should diversify their systems so they don’t depend on one country, one provider, or one AI model ecosystem. “For CIOs outside the US, Kubernetes and multivendor cloud environments will become increasingly important because they can support both local and global deployments,” the analysts wrote. HyperFRAME’s Walter said IT leaders should also tighten governance over the AI they already run. “Know which systems are in use, what they can access, what actions they can take, and who owns the result. Ask vendors for test results, incident information, and clear contractual responsibilities,” Walter said. Some things won’t change regardless of who controls Congress, Walter said. “Enterprises still have to establish whether AI produces better work after accounting for review, corrections, and operating costs,” she said.
Found in: Computerworld. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 1.