No zero-day was harmed in the making of this espionage campaign. The most consequential state intrusions of the moment did not require exotic exploits at all — they ran on unpatched routers, stolen credentials and patches that existed but were never installed. Nation-state espionage has entered its most aggressive phase in a decade, and alongside the break-ins runs a quieter programme: vast archives of encrypted traffic, stockpiled today for a quantum future that will make them readable. Two storylines, one strategic logic. Collect everything now. Exploit it forever.
Salt Typhoon burrowed in, and stayed
The anchor case is Salt Typhoon, the Chinese state-sponsored cluster that worked its way into telecommunications providers and other critical networks and then refused to leave. The official record is a 37-page joint advisory from CISA, the NSA, the FBI and international partners — "Countering Chinese State-Sponsored Actors Compromise of Networks Worldwide to Feed Global Espionage System," catalogued as AA25-239A and released on August 27, 2025. It says these actors have been operating since at least 2021, burrowing into telecommunications, government, transportation, lodging and military infrastructure worldwide.
Per the advisory, the objective is data that lets Chinese intelligence services track the communications and movements of targets globally. Not one ministry's inbox. Everyone's patterns, everywhere. The advisory also linked the activity to Chinese technology firms, including Sichuan Juxinhe Network Technology Co. Ltd., alleged to provide services to China's military and intelligence arms. Coverage of the advisory noted breaches affecting more than 200 U.S. firms alone, with ripple effects across dozens of countries.
And the method? Not zero-days. The actors are having "considerable success" exploiting publicly known, often unpatched CVEs — vulnerabilities in Cisco, Palo Alto Networks and Ivanti edge devices for which patches already existed. The door was not picked. It was left open, with the fix sitting in a vendor bulletin nobody applied.
Frankie Sclafani, director of cybersecurity enablement at Deepwatch, put the shift to Dark Reading: "Instead of just spying, groups like Salt Typhoon are now burrowing deep into critical infrastructure networks worldwide. This isn't just about stealing data; it's about gaining long-term access for potential disruption."
Telecom is not an arbitrary target. Carrier networks carry the metadata of modern life — who communicated with whom, from where, for how long — for millions of people at once, and they sit upstream of the governments, militaries and companies that rely on them. An adversary inside that layer doesn't need to breach each target one by one. Disclosure did not end it, either. CISA warned that Salt Typhoon was still lurking in telecom networks months after the initial cleanup efforts, and released follow-on hunting and mitigation guidance. Nor does the group sit alone: it belongs to a wider constellation of PRC-aligned clusters — Volt Typhoon, Flax Typhoon, Brass Typhoon — that collectively target not just espionage but pre-positioning inside critical infrastructure. Cleanup, in other words, is a claim you have to keep proving. An intruder discovered quickly is an incident to be cleaned. An intruder resident since 2021 has been reading the map for years.
Espionage with a time machine
Running parallel to the intrusions is a quieter programme with a longer horizon. "Harvest now, decrypt later" — HNDL — means adversaries intercept encrypted communications, financial transactions, medical records and government documents today, data they cannot read, archive it, and wait for quantum computers capable of running Shor's algorithm to unlock it years from now.
This is not speculation dressed up as strategy. The NSA, CISA and NIST have all publicly acknowledged that HNDL collection is already underway. Executive Order 14409, issued in June 2026, states that adversaries "may already be collecting" encrypted federal data. The 2013 Snowden disclosures documented mass collection of encrypted traffic more than a decade ago, and the NSA's Utah Data Center, completed in 2014, has been widely reported as built to store intercepted communications at precisely this scale.
The logic is unforgiving because storage is cheap and secrecy is long. The data most worth stealing has the longest secrecy lifetime: diplomatic cables, defence archives, health records, merger documents — all of it required to stay confidential for decades, well inside the window in which experts expect cryptographically relevant quantum computers. The Global Risk Institute's 2025 threat-timeline work suggests specialists see such machines arriving within roughly 15 years. And once harvested data is decrypted, the breach is retroactive and irreversible. There is no patching the past.
Here is the uncomfortable arithmetic: the defender's deadline is set by the adversary's archive. Data encrypted today under a 30-year secrecy requirement has to survive not today's computers but the machines of the 2040s. Which is why the EO 14409 dates — key establishment migrated by December 31, 2030, signatures by 2031 — are not bureaucratic milestones. They mark the point after which newly created secrets stop feeding the archive. The private sector has started treating them that way: Google set an internal PQC migration deadline of 2029 in early 2026, six years ahead of the NIST 2035 backstop, and Keeper Security began rolling out Kyber-based hybrid key encapsulation across its client applications in November 2025.
Industry voices have grown stark. On World Quantum Day 2026, e2e-assure's Gary Mounsor warned that a nation state could make decisive quantum progress before 2028, after which "Pandora's Box will be open" — predicting mass data dumps, credential abuse and "open season on corporate systems." Gigamon CISO Chaim Mazal is equally blunt: attackers are actively harvesting encrypted data in the expectation that quantum computers will make it readable, while 76% of organizations still assume their encrypted data is inherently safe. It is also why the migration itself has become an engineering discipline of its own, the subject of OmniWire's walk through the post-quantum migration roadmap built on NIST standards and Kyber-768.
Two stories, one doctrine
Salt Typhoon and HNDL look like separate threats. One is a present-tense intrusion; the other, a future-tense collection programme. Strategically they are the same doctrine applied across two time horizons: maximise the collection surface now, maximise exploitation options later. The telecom compromises harvest the metadata and content flowing through carrier backbones today. The quantum archives make sure today's encrypted sessions stay exploitable for decades.
That convergence is reshaping defensive priorities. Network defenders are told to prioritise the unglamorous fundamentals the Salt Typhoon advisory emphasises — patching edge devices, enforcing management-network isolation, detailed logging, coordinated threat hunting — because state actors are winning with known vulnerabilities, not exotic ones. The same pressure drives investment in faster detection across the SOC, since dwell time is precisely where collection happens: an intruder found in week one is an incident, an intruder found in year two is an archive. Simultaneously, CISOs have to treat cryptographic migration as an espionage countermeasure rather than a compliance exercise. The holes, when you look for them, are often embarrassingly basic — the same class of access failure that shows up in API authorization gaps that let attackers walk between tenants, and in the perimeter boxes that were supposed to be the strong point, as the still-unpatched edge flaws behind the FortiMail exploitation campaign keep demonstrating.
What defenders do on Monday
Start by hunting for persistence, not just intrusions. Salt Typhoon's lesson is that state actors establish long-term access through legitimate pathways — compromised routers, stolen TACACS+/RADIUS credentials, GRE and IPsec tunnels. Threat hunting should assume the adversary is already inside the management plane and work backwards from there. Then patch the edge like it matters, because the advisory's core message is that known CVEs on Cisco, Ivanti and Palo Alto devices are the attack path. Edge-device patching belongs at the top of the queue, ahead of the glamorous projects.
After that, inventory your cryptography and migrate by data lifetime. Build the cryptographic inventory that NIST, CISA and the NSA all recommend, and migrate data with 10-plus-year secrecy requirements first — that is precisely what HNDL targets. Adopt hybrid post-quantum key exchange now: X25519MLKEM768 is supported in current browsers, deploying it costs little, and it closes the harvest window for new traffic immediately. Finally, pressure-test incident response against state actors specifically, testing not just technical containment but executive decision-making around disclosure, law-enforcement coordination and customer communication under CISA's incident-reporting expectations.
One last uncomfortable point. Public attribution of campaigns like Salt Typhoon rests on classified intelligence that defenders never see, so enterprises are asked to defend against named adversaries whose capabilities are described only in general terms. The practical response is to stop optimising for named actors and start optimising for techniques. The advisory's value is not the name "Salt Typhoon" — it is the concrete TTPs: management-plane persistence, GRE/IPsec tunnelling, exploitation of specific edge-device CVEs. So convert every state-actor advisory into detection rules and configuration audits within weeks of release.
The asymmetry running through both halves of this story deserves a final statement. For the intruder, time is an asset: every month inside a carrier network adds collection, every month of delayed migration adds to the archive. For the defender, time is the cost — the harvest grows daily, and the window in which today's secrets can still be protected closes a little with each unpatched edge device and each un-migrated key exchange. The gap between the two sides is not capability; it is tempo. It is worth being precise about what would actually change the picture, because the list is short and none of it is mysterious. Edge devices patched on the vendor's schedule rather than the enterprise's convenience would close the campaign's main door. Management networks isolated as if they were the target — because, on the advisory's account, they are — would shrink what a foothold is worth. Cryptographic inventories, built before the migration deadlines force the issue, would let organisations move their longest-lived secrets behind hybrid key exchange first.
Harvest-now collection only pays off if the cryptography eventually breaks, which is why migration timelines matter today rather than in a decade. The practical counter is already written down: the post-quantum migration roadmap built on NIST standards, Kyber-768 and TLS 1.3. What is at stake when states skip that work is visible in Denmark's population-register breach, where 8.8 million records were taken — identity data that stays sensitive for the rest of those citizens' lives.
Sources
- Cybersecurity News — "CISA Publish Hunting and Mitigation Guide to Defend Networks from Chinese State-Sponsored Actors" (advisory AA25-239A): cybersecuritynews.com
- PYMNTS — "CISA, FBI, NSA Warn on China's Spy Network" (September 1, 2025): pymnts.com
- TechRadar — "US Government Says Salt Typhoon Still Lurks on Telecoms Networks" (CISA follow-on guidance): techradar.com
- TechRadar — "US Warns Chinese Tech Firms May Have Ties to Notorious Cyber Espionage Group" (NSA/CISA joint advisory): techradar.com
- The Quantum Insider — "What Is 'Harvest Now, Decrypt Later' and Why Should You Care?" (May 1, 2026): thequantuminsider.com
- Medium — "Experts Warn the Quantum Cyber Threat Is Three Years Away" (Mounsor, Mazal quotes): medium.com
- DCSA/CDSE — "Quantum Computing Threats to Cyber Security" job aid (CISA/NSA/NIST guidance): cdse.edu
