Kike Miralles

The minimum credible hardware to break real cryptography is now at the tens-of-thousands-of-qubits scale. Today's new paper from out-of-stealth Oratomic arrives at similar resource estimates as Iceberg Quantum's Pinnacle Architecture from a couple of months ago with a completely different platform. Atomic platforms can exploit their own advantages to match the potential performance of the solid state ones. But both papers also highlight how space-time trade-offs may become the key design variable for quantum hardware. The direct trade-off between qubit count and runtime means that counting qubits could quickly lose relevance as we reach bigger QPU sizes, and that different hardware platforms will likely pursue completely different paths to commercial scale. The catch (sorry) is that runtimes are still extremely long, magic states remain a big bottleneck, and even decoder performance will be a challenge. Integration challenges also loom large. Some serious engineering awaits!

@insane_analyst The man almost picked the peak down to the minute 💀

@julianweisser And like 90%+ of it can be solved by reading a single book so you have enough context to reason through things with your lawyers

Actually crazy how many founders enter legal agreements with investors without having any idea what hugely consequential terms mean.

@lithos_graphein Because if there's one thing we know for sure, is that money is always allocated on the basis of technical due diligence and logical expectations, lol

McKinsey: The shift in private investment for quantum indicates it's on the verge of becoming a commercial technology.

@zephyr_z9 We’re thinking of calling it SPU, which stands for Slop Processing Unit

BRUH 🤣🤣🤣🤣🤣🤣🤣🤣🤣

@bubbleboi So it’s cheaper to get you to grill the founders than to read the materials? Something something arbitrage

Btw to any VC’s who need me to do due diligence for their deep tech photonic bullshit startups I charge $1,000 an hour and if you want me to join a call and grill the founders I will do it for $500 an hour.

@martinmbauer You’re going to need some good reflexes to pull it off

Moving at 40% the speed of light, you’d only see the red button, but it'd look blue to you, while the real blue button is invisible You’d think you’re choosing blue like everyone else, when in reality you'd help sacrifice about half of humanity.. manslaughter via reference frame

@vin_sachi Isn’t there one trying to do this in upstate NY?

Will fund any co building actuators with fully American supply chain and manufacturing But first will need to see your CAD and your FEM/BEM solver

@arian_ghashghai Feels like a very long winded way of saying 1) this time is different and 2) you’re at a disadvantage. But to your point, that’s (mostly) only the case if your definition of VC is competing with a16z at their game with a $50M fund.

idk this essay relays a very articulate picture of the current zeitgeist in VC but also establishes a few opinions as fact that imo are empirically incorrect: The general premise here again is the now-common VC trope that only 10 companies matter, and everything else is chopped liver. I find this nonsensical because: 1) implies that only a company with certain input permutations (e.g. founder credentials) can grow to "matter", which enforces an elitism (e.g. "legible to institutional capital from day one") that only founders of a certain geography or milieu can matter (as we know, even this is wrong in the current climate). imo this is inherently the antithesis to the original "pirate" spirit that has historically driven VC returns (and why this industry even exists). also there is no empirical evidence that there are permutations of founder credentials that reliably predict outcomes 2) by extension of 1), that means everything priced cheaply (e.g. $5M post) is "lower quality" (and will not produce outcomes due to structural conditions). however, OP also says a way for small funds to navigate the narrow path is to invest in "opportunities in sectors or categories that are not yet legible to large institutional firms" and "to take bets where others won’t" (which is 100% correct btw). These verticals are exactly where those supposed poor $5M post deals are found btw (so the entire point on low price = low quality contradicts itself). 3) the assumption here is also that only companies hot on the private market will have liquidity opportunities (as IPO and M&A is "dead"). VC-backed tech co's not IPOing is primarily a product of cheap private capital, and frankly, brittle businesses that might have exploding revenues (which VCs value) but suck w.r.t free cash flow (which the public market values). incidentally, i would make the counter bet that this "narrow path" will actually produce *more* companies that are viable for the public market (by necessity, due to lack of access to cheap capital), but these might not be companies carrying the SF VC halo (OP uses OpenAI as the gold standard example, which is objectively a bad business in the truest sense of selling services or goods at a profit) fwiw I don't think Lucas is wrong per se. imo he's talking about a new category that's emerging as a product of AUM bloat in VC, which is still operating under the "VC" umbrella, but is not really VC anymore (and we should find a new name for it)

@yrechtman @ZacharyDeWitt In vs outbound

Best VC sourcing channels (in order): - Targetted cold outbound - Warm inbound from founder network - Warm inbound from other VCs - Cold inbound Curious on your take.

@IanRountree All the time. I’d say generally the “X minutes away” is off most of the time

*pays for priority uber because it’s 2 not 4 mins away* … “your uber will arrive in 5 minutes” 😑 Y’all have this same experience?

@blip_tm Nice flex

@caitlinbolnick1 If you don’t like my first principles, I have others 😂

Friend (who is a true operator) told me not to use the phrase "first principles" bc its an incredibly over-used VC phrase that therefore is meaningless and very few VCs are actually first principles thinkers. I've digested this feedback and I think it is very correct and true. So, I take back the use of the phrase 😂

Paper: arxiv.org/abs/2604.06319

So it's possible, after all, that no single qubit modality will rule them all. Please someone call me when we have a transducer layer that can make it work.

Quantum computing's future might look a lot like classical computing's past: heterogeneous. 🧵

Interesting paper showing time multiplexing for superconducting qubits may not be as bad as previously thought. Doing frequency multiplexing remains first route but issues with cross-talk may make some version of this attractive!

What ends up happening as you try to make your little excel model a bit more general each time