Google Cloud Sets 2027-2029 Deadlines to Quantum-Proof Its Encryption

Google Cloud laid out a formal timeline this week for retiring the encryption standards that protect most of the internet, betting that quantum computers powerful enough to break them will exist before the end of the decade.

The roadmap, published August 11, 2026, breaks the problem into three deadlines. By the end of 2027, Google Cloud wants to have neutralized what cryptographers call "harvest now, decrypt later" attacks, the practice of adversaries stealing encrypted data today and sitting on it until a quantum computer can crack it later. By end of 2028, the company aims to roll out quantum-safe digital signatures and certificates, plus the key-management systems needed to swap algorithms quickly if something breaks. Full post-quantum readiness across Google Cloud's infrastructure is set for 2029, a target the company first announced company-wide on March 25, 2026.

According to the roadmap, quantum-safe key exchange is already live on Google Cloud API endpoints and load balancers, using a hybrid approach that pairs the new ML-KEM algorithm with the older, battle-tested X25519 standard. The logic is simple: if researchers find a flaw in the new post-quantum math, the classical encryption underneath still holds the line. Google is also folding NIST-standardized algorithms, ML-KEM, ML-DSA, and SLH-DSA, into its Cloud Key Management Service. The National Institute of Standards and Technology finalized those standards in 2024 after years of public vetting.

Google isn't alone on this timeline. Both Cloudflare and Microsoft have set their own 2029 targets for post-quantum readiness, an industry consensus on when the clock runs out.

Why the urgency

The threat here isn't hypothetical paranoia. Encrypted data intercepted today—financial records, medical files, government communications, blockchain private keys—can be stored indefinitely. Once a sufficiently powerful quantum computer comes online, that data can be decrypted. That's the entire premise behind the 2027 deadline: whatever gets stolen now needs to already be protected by quantum-resistant encryption before it's even worth stealing.

Google's involvement in this space isn't new. The company has worked on post-quantum cryptography for roughly a decade and contributed research that fed directly into the NIST standards finalized in 2024. That's a legitimate case for taking Google's 2029 target seriously rather than dismissing it as marketing.

The hardware side of the race

While Google works on defense, the offense is advancing too. Quanta Computer, a Fortune Global 500 manufacturer based in Taipei, announced a collaborative development agreement on August 13, 2026, with Quantinuum, a trapped-ion quantum computing company headquartered in Broomfield, Colorado.

The deal focuses on building the physical infrastructure needed to manufacture quantum computers at scale, not just build one-off lab machines. Dr. Rajeeb Hazra, Quantinuum's president and CEO, said in the announcement that "it is time for quantum computing to transition from breakthroughs in physics achieved in the lab to breakthroughs in system manufacturing that can be deployed and operated at scale." Quantinuum says it has already commercially deployed multiple generations of trapped-ion systems using what it calls the QCCD architecture, claiming the industry's highest two-qubit gate fidelity based on its own published figures.

Quantinuum is not a small operation. The company says it employs roughly 700 people globally, more than 70% of its technical staff holding PhDs or master's degrees, with facilities in the U.S., U.K., Germany, Japan, Qatar, and Singapore.

The skeptic's case

Nobody actually knows when a cryptographically relevant quantum computer, one big enough to break current encryption, will exist. Estimates from researchers have ranged from a few years to a few decades, and they've been wrong before. Critics of the industry's urgency argue that spending billions to rebuild cryptographic infrastructure against a threat with no confirmed arrival date risks over-engineering a problem that may not materialize on schedule.

That's a reasonable objection. But it cuts both ways: the "harvest now, decrypt later" risk means the cost of waiting isn't zero even if quantum computers arrive late. Data stolen today doesn't need tomorrow's quantum computer to become worthless, it just needs one to eventually exist. And once it exists, encrypted data captured years earlier is exposed retroactively. That asymmetry—cheap to prepare now, catastrophic to be caught unprepared later—is the entire argument for Google, Microsoft, and Cloudflare all landing on the same 2029 deadline independently.

What's unresolved is whether that timeline holds. Google set its company-wide target in March 2026, less than five months before the detailed roadmap dropped. Whether the 2027 milestone for neutralizing harvest-now-decrypt-later attacks actually ships on schedule, and whether Quanta and Quantinuum's manufacturing partnership produces deployable hardware fast enough to matter, are both things nobody can verify yet. The industry has set the deadlines. Meeting them is a different question.