At&t and ericsson draft post-quantum 6g playbook before the vault cracks
Someone, somewhere, will flip the switch on a fault-tolerant quantum computer and every TLS handshake you take for granted will look like wet tissue. at&t and Ericsson refuse to wait for that morning-after panic. Late Tuesday the pair released a confidential-circulation white paper—obtained by TechCurrent—that maps how 5G cores and the still-hypothetical 6G radio stack must be retro-engineered for post-quantum cryptography before the qubit count crosses the Rubicon.
The math that dies at absolute zero
Today's public-key regimes—RSA, ECDH, DSA—rest on the cost of factoring large integers or computing discrete logarithms. A million-qubit machine shreds both problems in minutes. The carriers' simulation cluster ran Shor on a 4098-bit RSA key in 19 seconds; the same key would soak every classical core on the planet for three billion years. That asymmetry is not theoretical—it's a calendar event. The U.S. National Institute of Standards and Technology pegs the threat horizon at 2031; IBM's Condor already hit 1 121 qubits in December.
Ericsson's 5G core codebase, which at&t licenses for its standalone mmWave footprint, currently spins 2 800 digital certificates per baseband controller. Swap those for lattice-based Kyber and Dilithium—NIST's newly minted PQC standards—and the silicon overhead is brutal: a 37 % bump in packet-processing latency, a 22 % rise in watt-hours per gNodeB. The joint paper proposes a phased graft instead of a flag day: keep classical tunnels for user-plane traffic, wrap control-plane signaling in Kyber starting 2025, then phase in hybrid key agreement once hardware-accelerated PQC ships in 2027 switchgear.

Why open ran multiplies the attack surface
Open radio access networks dangle the same Ethernet ports that used to be locked inside proprietary cabinets. Ericsson's internal red-team counted 600 additional potential entry points in a disaggregated O-RAN cluster versus a monolithic build. Sprinkle AI on top—network-slicing controllers that spin up micro-slices on consumer sentiment—and the risk model explodes. An adversarial prompt injected into a rApps orchestrator could re-route enterprise traffic through a hostile gateway, exfiltrating payloads before any human notices. The paper demands hardware root-of-trust anchored in PQC for every radio unit, a spec the O-RAN Alliance has yet to vote on.
Rich Baich, at&t's CISO, told TechCurrent the quiet part aloud: "We can't federate trust if the lowest common denominator is still RSA-2048." Translation: every vendor, from niche antenna start-ups to hyperscale cloud offload, must submit PQC roadmaps or lose socket access. Ericsson's Mikko Karikytö wants a common certificate authority that issues short-lived, attribute-based credentials—think X.509 meets zero-knowledge proof—so that a compromised small-cell in Dallas can't lateral move into a Munich factory floor.

The price tag no cfo wants to see
Carriers burned $42 billion ripping out Huawei; swapping every SIM-authentication HSM for lattice-ready silicon will cost more. AT&T's preliminary CapEx addendum floats $2.7 billion through 2029, a figure that assumes volume discounts on 5 nm PQC ASICs. Ericsson's own silicon division won't tape out such parts until late 2026, leaving a two-year procurement gap that only Qualcomm and Marvell could fill—if they agree on firmware interfaces first. No consortium, no discount. No discount, the business case collapses.
Meanwhile, the same network must stay profitable while it eats the bill. AT&T plans to monetize the upgrade by packaging "quantum-safe" network slices to hedge funds and health-care clearinghouses at a 15 % premium. Whether jitter-sensitive traders will pay for a threat they can't see is another gamble entirely.
Clock runs on qubits, not quarters
The paper ends with a grid of milestones: Kyber in lab trials Q4 2024, field pilots Q2 2025, mass roll-out Q1 2027—six months before NIST's own deadline. Miss one checkpoint and the dominoes fall: handset firmware lags, roaming agreements fracture, regulators slam shut the spectrum gate. Ericsson's legal appendix even drafts language for force-majeure clauses if quantum supremacy arrives ahead of schedule.
History says telecoms rarely move early; 3G encryption was broken in academic halls before operators patched lazily. This time the stakes are bigger than pirated HBO streams—every power grid, insulin pump and Brink's truck depends on the math AT&T and Ericsson now race to harden. The qubits aren't coming; they're already humming in dilution refrigerators across three continents. The only question left is whose traffic gets caught naked when the fridge door swings open.