AnswersPublished18 sources
Can All Code Be Formally Verified by 2028? What Tenev’s Prediction Means
On November 4, 2026, Vlad Tenev predicted that all code would be formally verified within two years, starting with safety critical software. Formal verification can prove specified properties under a model and its assumptions; it does not prove that the specification captures every real world requirement.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- On November 4, 2026, Vlad Tenev predicted that all code would be formally verified within two years, starting with safety critical software.
- Formal verification can prove specified properties under a model and its assumptions; it does not prove that the specification captures every real world requirement.
- Tenev co founded Harmonic, which sells the Aristotle verification technology behind some of the examples, so his forecast is an interested prediction—not independent evidence of industry wide adoption.[14]
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What is Robinhood CEO Vlad Tenev’s November 4 prediction that all code will be formally verified within two years, starting with safety-crit. Article summary: Tenev’s November 4, 2026 prediction is that “all code will be formally verified within two years,” beginning with safety-critical software and spreading to less-critical code—effectively a deadline of November 2028.[16] . Topic tags: general, academic, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks
Vlad Tenev’s November 4, 2026 prediction is strikingly broad: all code will be formally verified within two years, with safety-critical software first and the practice spreading from there.16 That implies a deadline of November 2028. Harmonic’s recent examples show progress in AI-assisted verification, but they are limited demonstrations—not evidence that every codebase will be formally verified on that schedule.
What Tenev predicted—and why his role matters
Tenev had already said that “all software and hardware will be verified” after Harmonic announced a move into code and chip verification on October 27. The November 4 statement added a specific two-year timeline and a safety-critical-first sequence.15
16
Tenev co-founded Harmonic and is identified by the company as its executive chairman.14 That makes his prediction relevant to the potential market for Harmonic’s technology, but it also means it should be read as a commercially interested forecast rather than an independent assessment. Harmonic has attracted backing from Nvidia’s venture arm, Sequoia Capital and Ribbit Capital; a reported $120 million funding round last November valued the company at $1.45 billion.
31 Funding and valuation indicate investor commitment, not proof that the technology works for every use case or that adoption will meet Tenev’s deadline.
What formal verification proves—and what it doesn’t
Harmonic describes Aristotle as using formal verification to prove whether specified properties hold across the possible inputs covered by a defined system. That differs from testing, which checks selected cases, and from code review, which relies on people inspecting code.6
The distinction is useful, but a proof is only as meaningful as the property being proved and the system it describes. A proof can show that an implementation meets a specification without showing that the specification captures the real-world requirement. Models, preconditions and environmental assumptions also set boundaries on what the result covers. If a hazardous situation is excluded by an assumption, a valid proof may not address it.
So “verified” should not be taken to mean “flawless in every context.” Formal proofs can provide strong evidence about defined properties; they do not, by themselves, settle whether requirements are right, whether assumptions match deployment, or whether behavior outside the model is safe.
What Harmonic’s examples show so far
Harmonic announced on October 27 that Aristotle would extend from mathematical proofs to verification for C software and Verilog hardware.2
6 The launch is evidence that the company is applying its approach to engineering problems. It is not evidence of broad adoption across programming languages, organizations or software systems.
Some reported demonstrations are more specific:
- DIII-D reactor-control software: Reporting says the fusion laboratory used Aristotle to reconstruct part of its reactor-control software in four hours from one research paper and a 254-word prompt.
33 The result concerns part of the software; it does not establish that the entire control system was verified or that the reconstruction alone certifies safe operation.
- Ibex processor instructions: A research-watch report describes Harmonic’s November 3 work proving nine properties for a particular Ibex RISC-V multiplier/divider configuration against a Sail specification.
18 That is a bounded verification result, not proof of the full processor or every possible configuration.
- Scope of the checks: Harmonic’s announced work concerns specified properties in C and Verilog, rather than a general guarantee that a program or chip is correct in every respect.
6 The available reporting does not establish a complete, independently assessed account of supported checks, exclusions and limits across all use cases.
These examples matter because they show concrete attempts to apply formal methods to software and hardware. But demonstrations of selected code or configurations cannot establish that verification is practical, affordable and maintained across the full range of production systems.
What would make the two-year forecast credible?
The forecast is about widespread adoption, not simply whether a verifier can prove selected properties. Stronger evidence would include independently reproducible results, production deployments across varied systems, clear reporting on supported languages and toolchains, and evidence that proofs can be maintained as software changes.
Independent review remains important too. Teams still need to scrutinize requirements, models, assumptions and integration with the deployed system. Proof checking can strengthen engineering assurance, but it does not make those questions disappear.
The materials available here do not identify which, if any, components of Robinhood’s trading stack have been formally verified. That is an evidence gap, not proof that no such components exist.
For now, the most defensible reading is that Tenev has made an ambitious, commercially interested prediction, while Harmonic has reported promising but bounded examples. Those examples support the case for exploring AI-assisted formal verification; they do not establish that all code will be verified by November 2028.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Can All Code Be Formally Verified by 2028? What Tenev’s Prediction Means"?
On November 4, 2026, Vlad Tenev predicted that all code would be formally verified within two years, starting with safety critical software.
What are the key points to validate first?
On November 4, 2026, Vlad Tenev predicted that all code would be formally verified within two years, starting with safety critical software. Formal verification can prove specified properties under a model and its assumptions; it does not prove that the specification captures every real world requirement.
What should I do next in practice?
Tenev co founded Harmonic, which sells the Aristotle verification technology behind some of the examples, so his forecast is an interested prediction—not independent evidence of industry wide adoption.[14]