OrvalWintermute

I guess we should forget about:

• Larry Wu-tai Chin

• Peter Lee

• Chi Mak

• Katrina Leung

• Shujun Wang

• Fei Ye

• Ming Zhong

• Walter Lian-Heen Liew

• Kun Shan Chun

• Jinchao Wei

• Eileen Wang

All convicted of espionage

I want to echo the comments that you just made.

One of my primary methods of interacting with an iPhone is through speech and the state of Apple speech transcription is pretty horrible. It bothers me greatly.

I know some of the workarounds and things but it does feel like it’s in the Stone ages.

I don’t think it’s a microphone issue since iPhone microphones are fairly decent and I don’t think it’s a CPU issue either because Apple Silicon seems to be some of the best on the market. Which leaves us with the software…

Maybe they should put that cash hoard to good use and buy up some of these transcription companies or license their IP so we get truly high-quality transcription.

I think the company Taalas alone destroys Ed’s arguments

Because, comparing vs GPUs

~16k–17k tokens/second per user

<1ms latency

10x power efficiency

20x cheaper production

Model to Si ~ 60 to 90 days

We have every reason to believe SW_to_Si will facilitate improving economics

I'm increasingly thinking the same as our spend on tokens goes up.

If you have HPC or Supercompute already, you have much of the expertise on staff already to expand models locally, and between Apple Silicon and Exo there are some amazingly solutions out there.

Now, if only the rumors about Exo expanding to Nvidia are true..

More Industrial interior design than brutalist architecture

BigTech is quickly trying to punt their legal liabilities from their alleged actions, and transfer that risk elsewhere e.g. https://www.nbcnews.com/tech/social-media/jury-orders-meta-p...

Not sure this is truly the “Oracle VP of Workforce Economics” posting on X, but it makes a good thread:

https://x.com/gothburz/status/2040142920674656482?s=46

H1B is generally a giant scam of Labor Arbitrage

One of the most toxic TOS I have ever had the misfortune of reading.

Revocations works great in theory, and in theory & practice particularly in DOD.

The problem is a ton of certificate authorities consciously chose not to produce validation data previously, created insecure CAs, chose not to cache validation data, had knee jerk reactions to potential exposures, and many industries chose not to invest in technical capability to make revocation data available, performant, resilient, failing-over, failing gracefully, etc.

MITM is now the default for half the enterprise security solutions operating with cert to website “suspected good whitelists” which makes new domains on HN nigh unreadable

When your borders are porous, domestic surveillance becomes foreign surveillance

perhaps super efficient hybrids that blend the best of both worlds are really the future?

What about:

- native quick launch bar

- killing telemetry

- killing UI kludge

- permitting non-MS apps again

If you sell to the War Department, the CIA, the NSA, or ICE/Border Patrol, you know exactly how it is going to be used.

This after the fact naiveté by Anthropic is crazy.

Speaking as a patriot I’d be incredibly proud if my tool was used in a supporting role for one of the most perfect military operations ever executed.

This would likely be the Hallmark of my advertising for the near future.

Instead, we have anthropic going on a fishing expedition for information to claim a TOS violation and rightfully getting the boot! :)

Palantir was trying to do the right thing.

Anthropic is trying to sell to the government while simultaneously dictating terms on how AI is used in the war department.

When you sell to the War department, or the IC, you can be certain that they are going to use it for planning operations and for executing operations.

The story is they started fishing for classified info on how their tooling was used from their prime contractor, Palantir, who rightfully blew the whistle on them and told DOD what they were up to.

https://fortune.com/2026/03/07/pentagon-emil-michael-anthrop...

I support the rights of Democrat & Republican administrations to designate certain companies as supply chain risks OR a national security risk.

I do not want a TikTok hoovering up our personal info sending it to China

I do not want a Anthropic becoming essential to warfare then questioning when the their AI is used to bring an enemy to justice

I do not want Nvidia sending their latest and greatest to China

I do not want a ASML moving their super advanced photolithography machines to China

I do not want a DJI selling their drones in the US and then exporting all the meta-data back to China

I do not want a Huawei hacking through American IT companies and then getting a free pass on selling their devices based on stolen IP in the US

MS13 "Murder House" next door

Sure, No fire, no smoke.

> My conscious is a tiny bit cleaner

I think he meant my conscience.

Used to think Google was awesome when they were hyper accurate, fast, and not enshittifying products.

Now I am convinced they are just a little bit better than Meta.

Google has been evil for at least a decade, if not longer than that.

This is just pigslop masquerading as a moral stand.

What happened to the OG Google that cared about users, prioritized honest search, fast performance, and didn't murder pages with ads?

wow that is fast!

The RISC-V growth is fairly spectacular; moving forward the CAGR is estimated at > 40%

RISC-V could completely eat ARM for lunch if they try to jack up fees under the guise of "value extraction", and not through true value creation.

In 2 years from this date, I fully expect the safety critical RISC-V chips like the forthcoming High Performance Spaceflight Computer (HPSC) from Microchip, Inc.[0], and derivatives [1] [2] leveraging SiFive IP[3], and other RISC-V competitors [4] to take a dominant position in Space, Aerospace, Aviation, and potentially other less cost-sensitive industries where RTOS dominate.

This raises a few questions in my mind:

Could that extend to vehicles and other use-cases?

Will we see more derivatives with even higher performance beyond the already announced PolarFire2 designed specifically for terrestrial use?

I don't know how sensitive their overall BOMs are to high priced reliable chips designed for fault tolerance...

I don't know how fast the quality of the mass market chinese RISC-V chips will ascend the perceived quality gap, and expand offerings into newer profiles [5]

Where will the toolchain for RISC-V be on a specific chip basis?

Is Nvidia likely to expand their usage of RISC-V?

[0] https://www.microchip.com/en-us/products/microprocessors/64-...

[1] https://www.microchip.com/en-us/products/microprocessors/64-...

[2] https://www.microchip.com/en-us/products/microprocessors/64-...

[3] https://www.sifive.com/

[4] https://www.gaisler.com/products/gr765

[5] https://docs.riscv.org/reference/profiles/rva23/_attachments...

I love you!

I do talk with OSS devs about “we need X for security and we are willing to provide X amount of funding”

You’d be amazed how much OSS devs will do for you when your request of something they wanted to do anyways (but had no impetus for prioritization) is matched by a healthy rate

I feel much better after reading this because our organizations are: - funding OSS developers - engaging with OSS developers to determine potential funding priorities - providing project hardware at the project level - providing hardware to the individual OSS developer

While we are not to the point of hosting events in Hawaii yet, I’m hoping we can see this as a teaming arrangement to accomplish great things together!

Current state of the art Radhard & Rad Tolerant compute are way more expensive than terrestrial.

I agree, all the good papers definitely talk about custom designed radiators being used on the dark sides of data center in space.

This is written by someone that is not in aerospace that thinks terrestrially.

Engineering is always a question of tradeoffs.

Launch costs are dropping, and we’re still using inefficient rockets. Space elevators & space trains, among others, can drop this much more, the launch costs are still dropping, even using rockets, maybe we’ll never get to elevators & trains the costs will drop so low!

Radiation shielding is not required for VLEO or LEO, and phenomenally more capable aerospace processors are near - hi Microchip Inc! There are many other radiation solutions coming, no doubt with nuclear power.

Satellites can be upgraded at scale, though for many things, it does not make $ sense to upgrade them, but fuel , reaction wheels, solar panels, among other things do make $ sense to replace.

Latency was technically solved in 1995 & 2001 with the first laser comms missions NASDA’s ETS-VI kiku-6 and ESA’s Artemis , and Laser crossbars for comms are common. A full laser TDRS no RF is not yet extant but soon. Earth to deepspace was just demonstrated by ESA.

Cooling can be significantly improved due to lower launch costs, heat piping, RTGs, TEGs, and thermoradiative cells, not to mention sunside solar and darkside inline radiators

Furthermore, it is very likely that as neuromorphics with superior SWaP emerge, we could see very different models of space based computation.

Economic tradeoffs should drive many of these decisions as I’m not discussing the other applications of datacenter in space

maybe we just need License-Aware Vibe Coding that can link back to code snippet provenance similar to SBOMs?