通过现代数据架构释放AI潜力

Authors:Dawn Bushaus, Contributing AnalystMichelle Donegan, Contributing AnalystEditor:Ian Kemp, Managing Editor 3The big picture 9Section 1:What’s driving data architecture transformation?16Section 2:What needs to change in telco data architecture?22Section 3:What is a modern data architecture?28Section 4:The importance of data governance36Section 5:The role of standards in improving access to data41Section 6:CSPs start to modernize their data architectures We hope you enjoythe report and, mostimportantly, find ways touse the ideas, concepts andrecommendations detailedwithin. You can send yourfeedback to the editorial The big Earlier TM Forum research drew similar conclusions.In 2024, we polled AI decision-makers within CSPorganizations and fed the results into our GenerativeAI Maturity Interactive Tool, or GAMIT. Analysis of “There is an extremely close correlation betweenthose CSPs which scored highest in data readiness inour GAMIT survey and AI maturity,” says Mark Newman,TM Forum’s Chief Analyst. “Put bluntly, CSPs cannot While the initial survey and findings were carried outbetween June and the end of August 2024, garnering203 quality responses, GAMIT remained open fromAugust 2024 to March 2025. In that period morethan 1,000 individuals logged onto the tool, resultingin 408 quality responses in total. Of those, only 7% Democratizing dataTelcos’ big data strategies have evolved significantly Put bluntly, CSPscannot hope to deployGenAI at scale unlessor until they improve over the past couple of decades. With the rise of high-speed networks, smartphones and social media, datacomplexity increased exponentially, leaving traditionaldata warehouses unable to cope. Operators turned to Mark Newman,TM Forum But data lakes don’t have integrated governance,and while they excel at storing vast amounts of raw,unstructured data, they don’t provide optimizedaccess to it. That means that AI models have to AI needs the information and knowledge that comesfrom analyzing data. As one network architect noted inour survey: “Data without information is of little value.You need to have knowledge about the data to make CSPs are well aware that they need to democratizedata – that is, make it widely accessible and usableacross their organizations (and in the case of datamonetization, outside their organizations). And despitesignificant challenges they are making progress,particularly large operators. Throughout this report,we look at how leading companies are transforming nData is highly siloed and only accessible to a few departmentsnSome data sharing exists but significant barriers remainnData is widely accessible and integrated across most departmentsnData is fully democratized, with seamless accessand integration across the entire organization Managing and leveraging the petabytes of data thattelcos generate every day is necessary to becomeAI-driven, and this is a major goal for all the executiveswe interviewed. Their companies are embracing AI to Indeed, they are pushing towards highly autonomousnetworks that will rely on all types of AI, including What’s inside the report?In this report we look at how leading CSPs are building data platforms to modernize their approaches todata storage, optimization and orchestration. We lookat the drivers of change and evaluate the challengesCSPs are facing on their journeys. We also examine the Culture change However, modernizing data architecture is difficultbecause it requires operators to completely rethink howdata is stored, accessed, processed and governed. It’san engineering challenge certainly, one that requiresthe adoption of new methodologies like data mesh and The survey for the report was conducted early in2025. The infographic on the next page shows thedemographics of respondents. In addition to the survey, But all the executives we spoke with stressed thatthe biggest challenge is the requirement for afundamental shift in the mindset and culture of theentire organization. Telcos’ long history of buildingsystems that prioritize reliability and stability overinnovation necessarily has created a corporate culture Read the report to understand: •What it means to democratize data•The drivers for modernizing data architectures•Which data is most valuable•What telcos’ data architectures look like today•Which methodologies CSPs are using to modernizedata architectures Becoming AI-driven requires CSPs to view data as astrategic asset, which means embracing processesthat encourage cross-functional collaboration. Thetraditional keepers of data must give up controland collaborate with software developers, network Data is key to realizing meaningful financial gains Recent McKinsey & Company benchmarking of 20 large telcos globally highlights the importance ofdata, finding that most operators “have not built the data and AI foundations to achieve EBITDA impactat scale”. In other words, modernizing data architecture is imperative for CSPs to realize meaningful