S17 E41: Vik Sharma on Radar, Arbiter & Cake Wallet

Seven months ago, Vik Sharma joined Bitcoin Takeover Podcast and effectively declared the privacy wars over, telling the Monero and Zcash camps to lay down their memes because Cake Wallet would serve both. The peace, improbably, held. Now the Cake Wallet CEO returns with not one but two products, which is apparently what happens when a self-described workaholic perfectionist gets his hands on frontier AI models and a some spare time. If you haven’t already, listen to Vik’s previous appearance on the show too!

This episode is sponsored by Cake Wallet, Braiins Hashpower, Orange Rock, LayerTwo Labs, and SideShift. Cake Wallet also put up a $100 prize in Bitcoin, Monero, or Zcash for the livestream contest.

Radar: Signal, But With Lightning

The first product is Radar, a modified fork of the Signal client with Bitcoin Lightning payments built directly into the chat. Signal’s own attempt at payments, the ill-fated MobileCoin integration, never went anywhere, but it left behind payment rails that Radar could build on. The backend remains entirely Signal’s infrastructure, which Cake pays for with unprompted monthly donations, and a new linking feature lets you run Radar on your phone and Signal on your desktop without logging either one out. A native desktop version is in the works.

Why Signal over Telegram? Privacy, open source credentials, and usernames that keep your phone number out of the equation. Why Lightning over Monero, from a man considered a saint in the Monero community? Speed and reach. Payments on the go cannot wait twenty minutes, and Bitcoin still has the broadest appeal.

Monero and other coins may follow, but the real tell is what comes next: a Lightning-based stablecoin, possibly Spark’s USDB or Cake’s own, with optional auto-conversion so normies never have to watch their balance fluctuate. Vik dislikes stablecoins on principle and wants to ship one anyway, because Venmo users think in dollars. With over two million Cake Wallet users as built-in distribution, the usual stablecoin question of “who will ever use this” answers itself.

The AI Laboratory

The most revealing segment of the interview has nothing to do with crypto. Vik has assembled a private AI lab: two 512GB M3 Ultra Mac Studios running local models, and a growing fleet of Nvidia DGX Spark machines, eight at last count, bought slowly in case Nvidia drops a 256GB refresh. The motivation is characteristically blunt: his biggest fear is falling behind, and the only way to learn a technology is to get your hands on it.

The security argument for local inference is one every wallet developer should sit with. When Seth for Privacy (now Cake’s COO) ran local models through Arbiter’s codebase to hunt vulnerabilities, no third party ever saw the audit. Feed the same code into a cloud model and you are trusting that no employee, no log, and no gossip chain ever leaks what you were probing.

Vik also tells a story that will resonate with anyone who has fought AI guardrails: one frontier model refused to help fix issues in his own app, apparently suspecting him of hacking himself, while a second model complied without complaint. His conclusion, delivered to a Trump administration official over dinner in New York, is that America is repeating a mistake. He invokes a mid-nineties FCC event where Michael Powell argued it was too early to regulate the internet. The same, he says, applies to AI, especially when Chinese models are a prompt away and Italy’s grand ChatGPT ban amounted to blacklisting a single URL.

Arbiter: The Vibe-Coded Trading Terminal

The second product exists because Vik could not find a Hyperliquid mobile app he liked, so he built one without being able to read or write code. Arbiter is 100% vibe coded: he describes the outcome he wants, Claude asks clarifying questions, and a build arrives wirelessly on his iPhone. It has already processed over $5 million in trades on TestFlight, sports order types that Hyperliquid itself does not offer (validated by asking the AI to run 10,000 simulations), and ships as an APK, a Mac DMG, and soon possibly a Linux build.

The cleverest engineering is invisible. Because Arbiter runs on USDC via Arbitrum, Vik asked Claude whether deposits could be made gasless. Seven minutes of research later, the answer was yes: Circle’s CCTP handles cheap cross-chain USDC transfers, and a dashboard of funded gas wallets lets Arbiter pay users’ fees across Ethereum, Polygon, Avalanche, Base and Arbitrum. No more “go buy some ETH first.” When I pointed out that our sponsor Orange Rock, built by the Bitcoin.com crew, occupies similar territory, Vik’s competitive instinct fired instantly: we are gasless. Whether Apple and Google will allow a leveraged trading app into their stores is another matter, and geo-blocking the US is on the table.

Two caveats deserve emphasis. Everything AI-written that touched Cake Wallet itself, the Ledger multisig build, the Trezor integration, the shielded Zcash experiments, was treated as a learning exercise and deliberately not merged. And inside Cake proper, AI is used for auditing and review, not for writing financial code. The workaholic is also a cautious custodian.

Security After Coldcard

Vik donated $10,000 to the Bitcoin red team, and his reasoning cuts to the bone of the Coldcard RNG flaw that dominated Bitcoin Twitter. Everyone assumes open source means somebody smarter is watching. Coldcard proved that nobody was, for years, while scattered Reddit posts from users who lost funds were dismissed as user error. His prescription is unfashionably simple: passphrases. A 20 to 24 character passphrase carries more entropy than a 12-word seed and would have neutralized the entire class of failure, no casino dice required. Cake, he notes, is the only shielded Zcash wallet offering passphrases today.

The Privacy Narrative and the Multicoin World

On why privacy coins are outperforming, Vik separates the two horses he feeds. Monero has the usage; as the operator of the largest Monero wallet, Cake sees the real transaction volume. Zcash has the tech, at least until Monero’s full chain membership proofs land, plus the Grayscale trust, the Winklevoss twins, and treasury companies manufacturing momentum. His counterfactual is worth quoting: had a private coin launched first and Bitcoin second, everyone would ask why anyone would use the transparent chain.

His broader philosophy will annoy the maximalists in my mentions, and he says it plainly: being a Bitcoin maximalist is almost being anti-Bitcoin. The people who got in early were by definition the open-minded ones, and open-mindedness does not have a stop date. We live in a multicoin world, and the pragmatic privacy pattern of the moment, spending from shielded Zcash through NEAR Intents so the recipient sees only a liquidity pool, is still better than a fully transparent chain.

The Gatekeeper Answers

On listing criteria, Vik is disarmingly practical: demand, technology, and maintenance capacity. Kaspa has been on his watchlist for a year while its price slid. Dash is a genuine maybe, despite its own Twitter account once declaring it is not a privacy coin. Wownero and Haven were dropped when usage and upstream support dried up, and he reveals the quiet metric behind such decisions: one listed coin’s node was down for two months and not a single support ticket arrived. The counterexample is Exodus abandoning Monero over infrastructure headaches, with Cake migrating their users over. Privacy done properly is expensive to maintain, which is precisely why so few wallets bother.

Cake itself has crossed two million downloads, is adding roughly 3,000 installs a day, and is profitable on swaps, ramps, and Cake Pay, with growth accelerating over the past six months as the privacy narrative plays out. Next on the roadmap: a long-overdue desktop overhaul, more hardware wallet support, multisig, and eventually in-app Zcash governance voting. Expect to see him at Bitcoin Amsterdam in November, ankle surgery permitting, where I will be showing him my book on Bitcoin forks. Some of us still believe in physical media.

Listen to Vik Sharma on Spotify, Apple Podcasts, YouTube & more!

Watch on YouTube

Time stamps:

0:00 – Intro
1:18 – Vik Sharma Returns: Two New Products in Seven Months
2:41 – Radar: A Signal Fork with Bitcoin Lightning Built In
4:18 – How Radar Works: Signal Servers, Monthly Donations & Linked Devices
6:21 – Why Signal Over Telegram
7:36 – Why Lightning Over Monero
9:19 – Stablecoins for Normies: The Venmo-ification of Bitcoin
11:33 – Spark, Breez SDK & the Stablecoin Question
12:55 – The Nvidia Spark Cluster: Going All In on Local AI
16:25 – Why Local Models Beat Cloud Subscriptions for Security Work
18:03 – Guardrails: When Claude Refuses to Audit Your Own App
19:33 – Dinner with the Trump Administration: America’s AI Regulation Mistake
21:45 – Italy’s ChatGPT Ban & the Fear of Falling Behind
23:44 – The $100 Cake Wallet Contest
24:15 – Does Cake Wallet Use AI for Coding?
26:16 – Arbiter: A Vibe-Coded Hyperliquid Trading App
28:44 – $5 Million Traded, TestFlight & the Mac Desktop Build
31:13 – Linux Requests & the Orange Rock Comparison
32:17 – Gasless USDC: Circle CCTP & Frictionless Onboarding
36:10 – Can This UX Work on Private Chains? Zano, Aztec, Tari & Beam
38:44 – The $10,000 Red Team Donation & Cake’s Audit Results
40:34 – The Coldcard RNG Flaw & the “Someone Is Looking at the Code” Fallacy
43:14 – Passphrases: The Simplest Security Upgrade
44:58 – The Bitcoin Forks: Luke Dash Jr’s Blake2 & the Ecash Hard Fork
45:51 – Open-Mindedness: The Original Bitcoiner Virtue
47:44 – Wownero, MoneroV & Why Cake Dropped It
50:22 – The Winning Number Is 27 (and Nobody Picked It)
52:04 – The Privacy Narrative: Why Monero & Zcash Are Pumping
56:07 – “Bitcoin Maximalism Is Almost Anti-Bitcoin”
58:52 – NEAR Intents, Maya & Spending Shielded Zcash
1:00:53 – How to Try Arbiter Today
1:02:45 – What’s Next: Desktop Apps, Multisig & Back to Self-Custody Basics
1:04:26 – Still Bullish on Lightning
1:05:39 – The Gatekeeper Question: How Cake Decides Which Coins to List
1:10:07 – Dash Adopting Orchard While Zcash Moves to Ironwood
1:11:56 – The Maintenance Burden: Why Exodus Dropped Monero
1:13:25 – Zcash Voting, Sapling Addresses & Passphrase Support
1:17:03 – Seth’s Role & the Foundation Devices Connection
1:18:01 – Two Million Downloads & the Privacy Tipping Point
1:20:08 – Bitcoin Amsterdam & Conference Plans
1:22:14 – Sponsors & Goodbye