All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
We’ve built the first VPN protocol that breaks all speed barriers (Surfshark)
whynotlearn
Member
https://surfshark.com/blog/surfshark-launches-a-vpn-protocol
its a proprietary protocol and honestly some aspects of the blog post seems to be Ai generated as well namely
Unlike most VPN protocols, Dausos provides a dedicated tunnel for each user’s traffic. That means your data travels through its own clean, private path — separate from everyone else. No cross-traffic exposure, no interference, no shared overhead.
but that being said, if we take the article as face-value then:
Current VPN protocols weren’t designed for consumers. They were built for different purposes and later adapted for VPN use. That adaptation process introduces inefficiencies and limitations. We started with a clean slate. Our goal was simple: create a protocol that prioritizes what users care about most — security and performance — without compromise.
Dausos uses AEGIS-256X2 encryption, which delivers faster speeds than the industry-standard AES-GCM, especially on modern hardware. It also gives each user a dedicated traffic tunnel, eliminating the overhead and resource waste that’s common in adapted protocols.
The result? Better security, faster connections, and more efficient use of both server and device resources.
Dausos has a rare VPN industry feature — it is fully post-quantum resilient and delivers up to 30% faster speeds compared to other industry-standard protocols. Besides that it also differs in:
Could someone more explained within VPN terminologies help explain what sort of architecture of this supposed protocol could look like if its not vaporware.
Comments
How about speed of light barrier? /a
Im bored by it. Im wondering more what the supposed issue with wireguard and the obvious somewhat needed upgrade, depending on your location, amnezia is.
I do have a lot of clients that use both and they are absolutely not that tech-savvy, yet they had no issue getting it going.
Closed source >>>>TRASH
I dont trust anything from these vpn providers. The justification they gave for their new protocol is not even making me excited. They claim speed increase but my wireguard setup already reaches near 100% speed my isp offers. And post quantum protection is okay, I think when time comes even openvpn and wireguard will implement them aswell.
Physicians are interested in having a meeting with surfshark /s
Though to be fair, there are some really interesting things happening within the speed of information (optic cables) that I am aware of:
https://en.wikipedia.org/wiki/Photonic-crystal_fiber : from my understanding it is 30% faster than normal optic cable lines.
Japanese Break Bandwidth Record: 450 Tbps On A Single Standard Fibre Pair : https://subseacables.blogspot.com/2026/06/japanese-break-bandwidth-record-450.html
(it took me a lot of time to re-find the second source)
Also there are some experiments happening on calculations happening on light itself rather than bytes. I think its called photonics.
sometimes I wonder what might happen if both 1 and 2 are combined.
This garbage is just a little bit modified Wireguard.
When a provider locks you in with their proprietary software, it's better to avoid them.
you can not break physics no matter how much you try. And you can not achieve 100% optimum speed/transfer in any case. So there is a hard limit which won't be break ever.
Yeah wireguard is amazing. A lower hanging fruit than creating a new protocol like how surfshark has made is probably to use FPGA or hardware optimized wireguard. I was once into the rabbithole of FPGA and it seems to be a perfect use case of it (well combined with Jane street using it for sub millisecond trading and a guy on youtube using FPGA to control robotic hands and machines)
The thing that I am a bit interested upon is the audit.
I actually got so interested to read the PDF that they had linked https://surfshark.com/media/NOR-29-summary.pdf
From the pdf: > To ensure a transparent and collaborative engagement, Surfshark and Cure53 utilized a private, dedicated Slack channel for all communications. This space was open to all participating personnel from both organizations throughout the project
I am unsure how confident I am in reading a audit when it has to write about a private,dedicated Slack channel for all communications...
findings. Of these, seven were classified as security vulnerabilities and three as miscellaneous
issues or best-practice recommendations. It is important to highlight that the most severe
vulnerabilities identified during the audit were localized to the external hosting environment rather
than the Surfshark VPN Dausos protocol or its source code. Consequently, these were categorized
as out-of-scope (OOS) for the core protocol assessment. The remaining eight findings, all of which
were situated within the Dausos protocol’s scope, were rated at Medium severity or lower.
Surfshark team for their excellent project coordination, support, and assistance, both before and
during this assignment.
Not going to lie here, but could someone from the cybersecurity space tell me more about Cure53 because their audit lacked multiple details and it might actually lead to an bad impression on both surfshark and cure53. At the very least to be bigger than 2 pages for what its worth.
So I think what has happened seems to be that surfshark could still say on its website, hey look guys we have been audited and then the audit congrats surfshark and themselves on using a dedicated private slack channel. So private.
I don’t understand why any privacy-conscious person would choose a VPN provider with closed-source clients FFS. There are so many choices out there. How can you trust anything when the client is closed source.
A person who cares about privacy sometimes isn't well aware about how privacy works itself. We can't expect everyone to even know let alone care about wireguard. They just want to know that they are "right" and privacy-friendly. Sometimes they hear from youtubers about how spooky and unsafe the internet is and how VPN helps them and oh how convenient they are sponsored by NordVpn/surfsharkVPN
Though it does seem that there are people who are starting to get the gist of it and the choice of advice has fallen to two companies: ProtonVPN and MullvadVPN and perhaps tangentially IVPN or airvpn I don't think that there is one size fits all and even mentioning these two companies can lead to debate about which one is better or if there is true privacy at all.
And with that swathes of debates. Nothing really ends up happening because either the end user isn't on that forum or he doesn't wish to read 20 pages of debates.
So these companies like Nordvpn/surfsharkvpn basically sell the placebo pill of Privacy especially if they aren't using open source protocols.
To be honest, the placebo business is quite a lucrative one as they would hand any govt data that they might be asking for.
I think that there are only a very few companies if any, but I have heard of one or two which actively fight cases within court for what is essentially a 5$ per month service. I think that the best litmus test is to see which companies fight the cases in courts but those would also be the most unprofitable/unsustainable ones and probably more expensive as well.
Very skeptical just based on the quotes there. The first one I'd call marketing more than "AI". Wireguard, OpenVPN, simple socks proxies + socksifiers, all have their own "clean, private path" separate from everyone else. e.g, the instance may be shared, yet the end to end sockets / transmission isn't shared with anyone else...
There's no benefit to process forking (multi-processing) vs threading. There's also no real per-user performance difference between a multi-threaded design that uses non-blocking sockets with something like epoll or ioctl VS a blocking design, per client session, with threading or forking. From an efficiency standpoint, one process design is far more efficient / superior.
There are some things that you can do that actually have real-world benefit:
-- TCP splitting (reduce the ack latency and speeding up recovery time)
-- FEC & Interleaving (add extra parity to avoid re-transmission and add tolerance to impulse loss)
-- Forward TCP over "different" types of protocols such as KCP (faster retransmit, generally a more aggressive protocol without the exponential backoff)
-- Don't use protocols like TCP for the VPN iself, since nested force ordering & congestion control is detrimental to the contained sessions.
... but in perfect network conditions, TCP will perform the same as KCP, and FEC + interleave will just add latency.
Long distance link with loss? Yes, you can do better than just TCP over Wireguard. I've personally toyed with and benchmarked concepts like this many times. Unreleased projects, but may release them if I polish them sufficiently.
JIST: You're not going to beat Wireguard under most "normal" conditions.
Gemini please summary all the yap
I have suggestion for LET, introduce Premium subscription ($20/month) that adds AI summary button, and bundle a velox media VPS plan with it
Easy recurring revenue
If fairness is disregarded, employing UDP and QUIC protocols is evidently a better choice.
The creators of this new protocol set 4 requirements when they started working on it:
1) Break the speed of every protocol that there ever was ✓
2) Break the speed of every protocol that there is ✓
3) Break the speed of every protocol that there will ever be ✓
4) Break Speed Of Light ✓
All 4 requirements were met fully (✓✓✓✓,) this is the first truly Tachyon based Protocol ever developed and it is so fast that it makes information travel much faster that its carrier, it is truly superluminal. You can easily achieve speeds several times larger than the speed of light.
I've checked it and it is truly revolutionary, this new protocol bends and twists the laws of physics themselves. I'd say it's a bit like M$'s Majorana particles in Quantum Computing: REVOLUTIONARY!
Yeah, i'll admit that i haven't read any of the companies claims but i'll still call hot air by mere definition. There's a reason it took so long to get from OpenVPN to Wireguard. Encapsulating packets simply doesn't give the developer a lot to play with and once the more or less stupid shit is avoided performance differences will be marginal or at least highly context dependent. Someone claiming to have made major advancements shouldn't be surprised to be met with a lot of skepticism as long as they aren't willing to share their supposed magic formula.
Most of the "speed gains" of Wireguard aren't even the protocol and are just the migration from userland to the kernel. That anything in the kernel bypasses the socket kernel <-> user memory copies and thus is much lighter on cycles.
Don't get me wrong, there's alot that's good about Wireguard such as that the codebase is small, the protocol is simple, and it's much easier to audit than OpenVPN. But, comparatively like with OpenVPN w/ DCO, the protocol wasn't why it was faster.
Correct, there's no such thing as magic, and there's no such thing as free. All changes have a cost. Just like FEC isn't a magic bullet, nor is more aggressive retransmit, those have bandwidth overhead costs even if they 'speed things up' in the presence of loss or higher latency.
Similarly TCP store and forward / splitting consumes alot more memory vs just acting as a router, and you're changing the behavior in that connection is no longer end-to-end like some applications expect.
Add me to the list of people acutely disinterested (actually anti-interested) in surfshark's snake oil "racing VPN" thingy.
@a2razor already provided some wake-up coffee. To be frank though I'm under the impression that OP doesn't have the first clue what he's talking about and seems to approach the topic like a visit to Disneyland, sorry.
I happen to know a thing or two about AEAD as I in fact had discovered a grave error in one of the CAESAR finalists reference implementation, grave as in "no need even to hack".
What disturbs me the most personally is that surfshark seem to not really have understood the field and such passed/ignored an excellent performance enhancement opportunity. Short 30000 feet explanation: THE relevant here difference between TCP and UDP boils down to assured packet delivery. And AEAD offers even two opportunities to "get TCP assurance with UDP" - and they recognized neither of those (using the nonce is the first one and the 'AD' in AEAD is the second one).
Another point is that one can have more speed than needed because speed doesn't come for free. The third point is that AEGIS basically stole a real cryptologist's design and just slapped their name on it.
Yet another and IMO decisive point is that a very well known and highly regarded AEAD solution by another real cryptologist is available since more than a decade and has been subjected to extensive cryptanalysis. This Aegis thingy though is from 2025. Translation: battle-proven? NOPE. Besides, the version they chose has a state space of 768 bytes, which means that it doesn't fit in the L1 cache.
And btw, one of the two heroes describes himself as "an entry level SOC designer" and the other is some employee of Fastly.
TL;DR FUCK the new surfshark toy!
(I am btw a happy surfshark customer ~ not biased against them)