AT&T CTO Details 5G and Fiber Expansion Plans, AI Network Push, and Satellite Direct-to-Cell Outlook

AT&T (NYSE:T) Chief Technology Officer Yigal Elbaz outlined the company’s network and technology priorities at a conference discussion held virtually, emphasizing a customer-first approach, ongoing wireless modernization, rapid fiber buildout, and increased use of artificial intelligence in network operations. Elbaz also addressed how AT&T is thinking about fixed wireless access, AI-driven changes in network traffic patterns, long-term industry shifts around “6G,” and the role of satellite-based direct-to-cell (D2;C) services.

Customer experience, reliability, and modernization

Elbaz said AT&T views network priorities through the “lens of our customers,” starting with basic connectivity and consistency of experience. He pointed to dropped calls as an example of reliability, stating that AT&T has “the lowest dropped call rate in the U.S.” He also highlighted AT&T’s role supporting first responders through operation of the FirstNet network, which he said raises the reliability bar even higher.

On capacity, Elbaz tied demand growth to connectivity needs as well as AI and video traffic. He said AT&T’s wireless modernization includes putting its mid-band spectrum to work, simplifying mid-band spectrum and cell-site configurations, and “opening up the architecture” to improve efficiency and reduce costs. He added that AI can help solve what he described as an optimization problem—balancing customer experience, devices, reliability, coverage, capacity, and use cases to guide proportional investment decisions.

Two scaled platforms: 5G wireless and multi-gig fiber

Elbaz described AT&T as building “two scaled network platform[s]”: a modern, open, nationwide 5G wireless network and a “metropolitan multi-gig fiber network.” He said the fiber network already covers 36 million living units and claimed no one else has a fiber network of that size or is building at AT&T’s pace.

He framed the benefits of operating both networks in two ways:

  • “Owner economics” flexibility from building and operating both wireless and fiber assets
  • Support for AT&T’s convergence strategy, including running multiple access technologies off a fiber foundation

Elbaz said AT&T is also changing how its services converge in the network. Historically, he said, services converged “deep in our core backbone” using different architectures and infrastructure. Under modernization efforts, he said AT&T is “pushing out all of that convergence to the edge of our network, to each one of our central offices,” with services running on a “converged and intelligent architecture and infrastructure.”

How fixed wireless fits alongside fiber

When asked how AT&T prioritizes fiber, fixed wireless access (FWA), and wireless, Elbaz said fiber is the “lead offer” in areas where AT&T has fiber assets, arguing that “fiber is the best broadband technology” because it is symmetrical and reliable.

He also laid out scenarios where fixed wireless plays a role, including:

  • Areas where AT&T does not yet have fiber assets
  • Neighborhoods where fiber expansion is planned but will take “a year or two”
  • Use as a “catch product” to move some customers, including small and medium businesses, off older infrastructure
  • Customer segments that prefer other service types

Elbaz said AT&T recently gained additional spectrum and, “in a matter of weeks,” deployed additional 3.45 GHz spectrum across 23,000 cell sites through what he described as a short-term spectrum management lease. He said the added capacity expanded geographies and increased the addressable market for fixed wireless capabilities.

AI’s impact on traffic patterns and enterprise demand

On AI-related network demand, Elbaz said AT&T is not yet seeing a “completely different” mode of network operation, but it is seeing “signals.” He pointed to consumer use of AI on smartphones and the potential for AI wearables, noting that these applications can be more “upstream” heavy. He cited examples such as autonomous cars or robotaxis, saying that during training phases the upstream ratio can exceed 50%, which he characterized as a shift from historical patterns.

Elbaz also said AT&T is seeing growing demand for high-speed, low-latency services—specifically “400G”—in the enterprise space. He discussed “agentic AI” and suggested agents could drive more continuous, unpredictable traffic patterns, including more “east-west” traffic compared with today’s behaviors. He also cited “physical AI” use cases such as autonomous cars, drones, and humanoid robots as drivers of new network requirements.

Elbaz said AT&T has assets he believes position it to be “AI-ready,” including low-band spectrum—specifically noting 600 MHz as useful for upstream wireless services—an open architecture that can adapt to usage patterns, symmetric fiber broadband, and metro investments intended to support 400G and above over time. He added that AT&T has fiber connections to 600 data centers in the U.S., which he said helps customers run AI workloads in different locations with high-performing networking.

Internal AI use, 6G, and satellite D2C

Elbaz said AT&T established a Chief Data Office more than a decade ago and has progressed from big data to machine learning and AI, later expanding into generative AI capabilities at scale. He said the company is applying AI across functions including software development, customer service, fraud detection, and finance.

In network operations, Elbaz said AT&T has used machine learning and AI for years and that improved AI capabilities and movement to open architecture have accelerated progress. He cited examples including “cell site sleep” for energy savings, claiming newer models can improve efficiency by 20% to 30%, as well as self-optimizing networks and a “3D ray-tracing model” to better understand cell-site propagation. He said these capabilities can help the network adapt to disruptions such as fiber cuts or weather events and readjust in near real time to avoid customer impact.

On 6G, Elbaz argued the industry should “decouple the Gs” from innovation cycles, contending that software, openness, AI, and cloud-native infrastructure do not evolve in five- or ten-year increments. He said AT&T’s Open RAN strategy is aligned with continuous innovation.

Regarding edge compute, Elbaz questioned the value of pushing compute “all the way to the far edge” in the U.S. merely to save one or two milliseconds of latency, citing what he described as extensive data center investment and the proliferation of high-performing compute in metros. He said AT&T’s opportunity is to provide deterministic connectivity that links use cases to the appropriate models and infrastructure distributed across the country.

Finally, Elbaz said satellite direct-to-cell service is a major industry innovation and described it as complementary to terrestrial coverage. He referenced AT&T’s partnership with AST SpaceMobile and said D2C can help provide service in areas terrestrial networks cannot cover. While he said D2C is valuable, he added he does not see it becoming a “true competitive offering” to terrestrial service in the near term, citing limitations around beam creation and indoor experience.

About AT&T (NYSE:T)

AT&T Inc is a global telecommunications company that provides a broad range of communications and digital entertainment services. Its core activities include consumer and business wireless services, broadband and fiber internet, and network infrastructure. The company operates branded wireless services through AT&T Mobility and deploys fixed-line and fiber networks to deliver high-speed internet and related home services.

AT&T’s product and service portfolio spans mobile voice and data plans, smartphones and device sales, home internet (including fiber-to-the-home where available), and managed connectivity solutions for enterprise customers.

Read More