Signal · Volume I · February 2026
The Infrastructure
Argument
Twelve essays from the engineers reshaping connectivity.
Open RAN migration paths. Spectrum auction fallout. Fiber overbuilder economics. Arguments dense enough to cite in a slide deck.
Dr. Elena Vasquez
Chief Network Architect · Meridian Telecom Group
Fifteen years designing RAN infrastructure across three continents. Former spectrum advisor to the FCC's Broadband Task Force.
Essay
Open RAN's Interoperability Debt
The O-RAN Alliance's disaggregation promise conceals a $4.2B integration liability that no vendor roadmap acknowledges.
When a Tier-1 carrier commits to Open RAN, the public announcement describes savings. The internal architecture review describes something else entirely — a sprawling matrix of interface specifications, each versioned independently, each carrying the quiet assumption that someone else will resolve the conflict at the seam. The xApp ecosystem alone has produced 340 registered interfaces in 18 months. Fourteen are deprecated. Sixty-one have no certified implementation. The remaining 265 are "under validation" — a phrase that, in network engineering, means the problem is real and the timeline is not.
"Every disaggregated interface is a contract without a counterparty."
Chukwuemeka Okafor
Director of Spectrum Policy · Continental Infrastructure Partners
Former ITU-R delegate. Shaped the 3.5 GHz CBRS framework. Now advising private equity on spectrum portfolio construction.
Essay
The Auction Arbitrage
C-band repack has created a 180-day window in which secondary-market spectrum licenses are systematically mispriced by 30–45%.
The FCC's C-band transition proceeded on schedule. The interference coordination did not. In the 14 months following auction close, 23 incumbent earth station operators filed for extended protection periods, each filing effectively freezing adjacent license utilization in corridors that account for 11% of the addressable 5G mid-band market. The secondary market has not priced this. Licenses in affected markets trade at par with unencumbered equivalents. The arbitrage window is narrow — regulatory resolution is expected in Q3 — but the spread, for a patient holder, is approximately 38 cents on the dollar.
"The market is pricing spectrum as if the interference problem solved itself."
Astrid Lindqvist
VP Infrastructure Strategy · Nordic Fiber Holdings
Built and sold three fiber overbuilders across Scandinavia and the Baltic corridor. Now sizing US markets for institutional capital.
Essay
The Overbuilder's Dilemma
Fiber overbuilders in Tier-2 markets are generating IRRs that pencil at 12–14% on paper and deliver 7–8% at exit — and the gap is always the same two line items.
The pro forma for a Tier-2 fiber overbuilder is a work of optimistic engineering. The take-rate assumptions are derived from markets where the incumbent's product is genuinely inferior — which is true in approximately 60% of target markets. The construction cost assumptions are derived from the first 40% of the build, before the geology changes and the permitting office discovers it has opinions. The two line items that compress every return are make-ready costs and customer acquisition cost in year two, when the promotional pricing expires and the incumbent responds. Neither appears at full weight in any deck I have reviewed from a first-time overbuilder.
"The make-ready estimate is where the return goes to die, quietly, in a footnote."
Rajiv Mehrotra
Principal, Telecom Infrastructure · Blackstone Digital Assets
Structured $6.4B in tower sale-leaseback transactions. Leads private equity evaluation of neutral-host and shared infrastructure plays.
Essay
Neutral Host's Valuation Ceiling
Neutral-host indoor DAS assets are being valued on EBITDA multiples borrowed from tower REITs — a category error that will correct sharply when anchor tenant churn becomes visible.
The tower REIT multiple is a function of contractual certainty. An anchor tenant on a macro tower signs a 10-year lease with three 5-year extensions and pays an escalator regardless of traffic volume. A neutral-host DAS anchor tenant signs a 5-year agreement with a 90-day termination clause and pays on a per-connection basis. These are not the same asset. The market is currently pricing them within 1.2 turns of each other. The correction mechanism will be the first major enterprise anchor that declines renewal in favor of private 5G — an event that, based on current RFP volume, is approximately 18 months away in at least three top-10 US markets.
"You cannot borrow a tower multiple for an asset with a 90-day exit."
Dr. Wei-Lin Chen
Head of Regulatory Affairs · Pacific Spectrum Advisors
Former senior counsel, NTIA. Negotiated the FirstNet spectrum transfer. Advises carriers on CBRS, mmWave, and mid-band strategy.
Essay
The NTIA Funding Trap
BEAD program recipients are structurally incentivized to overbuild areas that do not need fiber and underlevel areas that do — and the challenge process will not catch it in time.
The BEAD challenge process was designed to prevent subsidy capture by incumbents claiming serviceable areas that are not, in practice, served. It functions as designed for that narrow purpose. It does not function for the inverse problem: applicants who have mapped unserved areas with sufficient accuracy to win funding, but whose construction timelines assume permitting velocity that no state has demonstrated. The 42-month average from BEAD award to first subscriber connection, based on comparable programs, exceeds the program's implicit model by 14 months. In three states, the mismatch is structural — the permitting authority does not have the staffing to process the volume the program requires.
"BEAD is funding the map, not the territory — and the territory has a permitting office."
Marcus Ostrowski
CTO · Ironclad Network Solutions
Designed the RAN architecture for two greenfield 5G deployments. Previously at Ericsson's Advanced Technology division for 11 years.
Essay
The Private 5G Reckoning
Enterprise private 5G deployments are failing at a 61% rate in year two — not because the technology is wrong, but because the integration layer was never scoped.
The private 5G sale closes on the radio layer. The deployment fails on the integration layer. Every enterprise IT environment has an OT network that was designed before 5G existed, a security perimeter that was not designed for a new radio access point, and an operations team that was not trained on a system that behaves differently from Wi-Fi in ways that are not intuitive. The 61% failure rate — defined as a deployment that is either abandoned, significantly descoped, or fails to achieve the business case within 24 months — is not a technology failure. It is a scoping failure that the industry has a structural incentive not to discuss, because discussing it would slow the sale.
"The private 5G sale closes on the radio. The deployment fails on the integration layer no one scoped."