Not the best analogy; water is famously not wet; it makes a surface wet is the common consensus ; wetness is being a property of a surface
4,953 karma · joined December 9, 2015
Exercising your vested options is a taxable event[2] and requires you to upfront some cash to buy the stock depending on what price(and rounds) your options were issued at- this price usually is considerably less than enterprise value of the same share lets say it is $1 and FMV of $100 at the time of issue, however even just $1 for vested 500,000 options is still $500,000 now.
For this simple[4] scenario you need $500,000 immediately to exercise the options into stock before they expire and about ~$13.86M by April to own[5]all stock if ends up say trading at $200 you are worth $100M then.
There are all kinds of funds who would loan you the money or buy the stock off you now on the secondary market, but vested options does not mean owned stock, and selling quickly after exercise is much higher tax under some scenarios.
----
[1]Some companies have generous(relatively) terms ex-employees to exercise up-to 5-10years later; but it not default in the industry.
[2] This close to IPO that should be relatively small risk; i.e. if the company pulls IPO you still have to pay tax end of the year; if you exercise them.
[3] NSO/ISOs work quite differently, and options vest monthly in most corps and there are lot of creative structures out there; so cannot say how it would be for ANT employee specifically.
[4] Simple - because there usually more than singe 409A valuation and more than single set of options plan issued and plenty of other complex calculus only a finance expert would advise on.
[5] You could sell the shares to pay the ~14M-ish in tax but that could be disqualifying sale(2 year after grant, 1 year after exercise) for ISO meaning you will need to pay effectively 40M tax on your 100M holdings versus only ~$27M otherwise, so you would a take loan against the shares rather than sell to pay the tax.
all training runs have to at some level share the compute capacity ; while they have more than enough to run many large runs concurrently, they don’t have enough compute to run everything concurrently.
> backed by Google Cloud’s data governance and high-performance infrastructure
Self-service retail buyers of cloud have access to some of these price advantages as well, 1-3 Year reservations(even paid monthly) are 40% cheaper of the bat after all. PAYG with fluctuating usage is quite more expensive than committed consumption for cloud.
Meaning they can swap out compute/storage spends on GCP to LLM model spend or use left over budget ( which would go waste anyway ) and so on . Any OpenAI sales by GCP teams would be considered against their quota typically in this type of arrangement [2]
This is normally not done to third party[1] marketplace services, Azure already does this for OpenAI, it helps drive some enterprise spends as it is from for fungible budgets or different approving authority internally in a org when it is not a new vendor, new budget etc.
----
[1] It is not too hard to get into these types of programs i.e. it not only for OpenAI level partners mid sized startups can also be part of these deals in marketplace, the requirement usually is minimum size and the partner having to prove/show that are spending that customer's revenue on the compute etc so money is going back to them, then AWS/Azure/GCP will consider an org for this type of program.
Given that GCP is hosting these models on their cloud and therefore OpenAI is spending a % of every dollar back on GCP for every token sold like this; it wouldn't have been too hard for OpenAI to got such a program even if they were a mid sized startup.
[2] Multiples maybe involved (also used to push some first party products) 1x would 1M in OAI sales counts as 1M in their Qtr/annual targets 0.5x would mean 500k and 5x means 5M quota would be retired for 1M in sales etc.
That is exactly my point, Human life supporting environments are great for tourism or experimental research missions where return on investment is not a critical factor.
Sending humans is not economically scalable option and cannot be ever cheaper than small carbon (or silicon based) based bots/organisms are the ones who can actually spread some version of life in the solar system and beyond.
---
This is same dynamics in fighter plane or submarines at smaller scale. We will get an interregnum period where the drones would be pitched that multiply/augment the ability of the pilot as in some proposed 6th generation designs.
This only serves mostly to get to the current crop of pilots and leadership to sign-up on a new program, 7th generation will be purely only drones. The full switch will happen when some rival Air Force will develop/skip to drone only 7th gen and the U.S. would have to catch up if not lead it. Submarines are also in the similar trajectory.
If it means life , then We don’t need humans to make life interplanetary at all; that is least efficient way to go about it ; we should instead design/choose lifeforms that are most likely to survive the target environment not depend on brittle/expensive life support technology to spread life through the system.
If the you is intelligence , then spreading AI could be a better approach rather than humans who are unlikely to be able evolve to survive in a hostile environment naturally anytime soon.
Humans in our current form in space is always going to expensive tourism no matter how much science fiction likes to pretend otherwise .
Nobody would be concerned if they didn’t have access to curl and used wget instead and didn’t ask permission .
However if that solution involved breaking the law yes I absolutely do not want a representative of me breaking the law and yes it is breaking the law whether we like or not under CFAA.
It is not about finding solutions to problems on their own; the solutions should respect the agreements we have made and the law .
The WIPE/privacy extension developer is using a pre-existing(hard-coded or generated) fingerprint to “mask” the user from meta .
Meta implemented a quality of life feature powered by the ability to set cookies and track you across the web of course and help users with forgotten accounts to unify them - but not a security issue you cannot do anything unless you are proving access governed by standard auth protocols ; the extension developer’s hack to mask created a broken experience because of the reused IDs.
Adblockers can create unintended effects , it is hard to reverse engineer complex platforms like meta and also hard to keep up to date with billion dollar companies code updates by small indie teams of volunteers ; doesn’t mean we shouldn’t use them, just be aware that it may happen when enabled
Last year Github said they are giving away 11.5B action minutes [1] for public repos for free, that translates to about ~30,000 cores . The explosion in commits, pull requests and action minutes that Vlad the CTO mentioned[2] in the August postmortem is approximately 10x and he also highlighted they added 3 Million cores to their fleet.
Assuming a similar 10X growth in open source; total public repo compute budget is only 10% of the new compute they have added. Only half of Github is on Azure as of April, and Azure itself is much much larger.
Last year numbers were worth $180 Million going by list prices of action minutes [3]. However that would be only $20-30M equivalent typical outlay for a mid-size tech company if they were actually buying 30,000 cores on the cloud ; and definitely much cheaper for Azure's procurement ;
It is a good PR strategy and very good deal for Open Source but is really only a small customer acquisition line item for them and doesn't likely meaningfully impact their compute issues one way or other.
[1] https://github.com/resources/insights/2026-pricing-changes-f...
[2] https://github.blog/news-insights/company-news/the-august-17...
[3] Action minutes are priced more expensive than even buying on-demand compute on the cloud including with Azure;
There are also third party providers of IP annotations to classify known VPN address ranges that content providers typically subscribe to blanket block providers.
The reason for this aggressive approach is streaming apps all need your IP as core signal for tagging your region and all content licensing is region locked (even on YT).
Netflix are/were the most relaxed about it , and for long time would only buy content if they got global distribution rights, but not anymore. Many VPN ads specifically used to market that you can watch Netflix geolocked content.
[1] IME they block DC IPs too although not needed for blocking professional VPN, even self hosted OpenVPN on cloud box usually gets flagged.
Sub-agents not sharing context is a useful design-pattern when you want adversarial or independent reviews.
Cache reads could be shared between sub-agents, A single node(8GPU cluster) supports few hundred concurrent user sessions, that all share the same KV cache memory, so it is likely model providers do colocate your sub-agents in one node, it is more efficient , but may not be guaranteed so performance could vary; like we have with elastic compute and storage[1]
This can be cheaper depending on your coding flow i.e. cache hit % and the billing plan - cache reads are basically free or charged very little in subscription plans.
[1] Modern AWS does offer collocation at additional costs for compute but that is not the default and most other clouds do not offer it
Middle East runs on poor people from the subcontinent and Africa.
Hong Kong depends on live-in foreign labor who have to camp out in public spaces on their day-off.
America is built on poorer labor of every kind - enslaved African cotton farmers , indentured Chinese railway workers, famine stricken Irish, or more recently Latin American or H1B Indians in tech.
Some countries try to attract the richest percent like New Zealand, however it is not economically meaningful nor is major portion of immigrants globally or within their countries even.
Countries may not like foreign poor immigrants, but they certainly need them to work harder for less pay and rights.
Nobody is saying nuclear is cheap it is not - we will be building a plant once in while for pure strategic reason no matter the cost anyway. The point is construction cost comes down per MW with higher frequency of plant building, and this is the case SMRs, we can get cheaper construction both Per MW and overall TCO in absolute dollar terms relative to large plants with SMRs.
Flameville 3 is the first reactor for France in 21st century and suffers the frequency problem I am arguing. It was a path finder therefore only one was ever ordered. EPR2 is 6 under construction and France does have a track record of doing it consistently at scale for cheap(by nuclear standards).
The Asian giants all do it for cheaper because they build regularly, Japan (post Fukushima policy direction notwithstanding) and South Korea achieve overnight numbers of $5-6M/MW or less, so does China.
Regular construction is key, to the industry having a chance, without SMR that is never realistic anymore.
AP1000 (reactor as Vogtle 3/4) in Taishan is estimated to cost $4M-6M /MW compared to $15M/MW+ at Vogtle. if Chinese safety standards seem lax, South Korea does regularly reactors at $4-6M/MW range.
Even Flameville 3 in France was significantly cheaper than Hinkley point C (same reactor type), which infamously required some quite ineffective Fish speakers. The EPR2 series (6 reactors) in France is expected to cost considerably less than Flameville did.
Key to cheap nuclear - SMR or otherwise is regular construction. SMR makes regular construction more likely as each plant is not a $10-20B+ capex that is a tough sell to tax payers particularly when it inevitably overruns, SMR also potentially have a bit less regulatory hoops to jump through due to their smaller size.
Not sure how that would work with building say a docker image, reproducible builds are pretty hard problem to solve, and caching intermediate layers is not always simple or even doable, we typically still need to publish to a registry which is not the cache server.
[1] https://googleappengine.blogspot.com/2008/04/introducing-goo...
The final asset/artifact is rarely small either. even best optimized artifacts can be few hundred MB docker image or more commonly multiple image layers running GBs in size .
each step is a network pull then recompute cache if stale and keep going till end .
Even fully cached outputs needs to fetched and read from a remote server[1]. A step n-1 outout fetched from remote cache server need to written to disk and then again read by step n[3] - all disk I/O and network bound operations.
10s may be achievable/realistic goal in the Java/C++ world where Bazel normally seen. In TS eco-system most people would be over the moon to get into ballpark of 1-2m for a decently large monorepo.
We should define Build more clearly here, if you mean running just transpile/compile steps or the full series of steps that includes tests (as the linear post here is talking about). It is hard to see even a small sub-set of a large suite of test that require a virtual DOM or a real browser can run in 10s or less.
[1] Typical for say managed CI setup .
[3] Common run-of-the-mill frontend + backend stacks in different languages etc.
We don’t complain about switch cases in code when there is two or three switches we start refactoring once it starts to proliferate.
The law is no different , passing a wider ban would not get the votes easily or quickly and the interested parties the onion industry have no reason to push for it neither does the lawmaker acting on their interests.
If said small markets also had exceptions passed seeing the onion one there could have been case to be broad.
It would premature optimization to otherwise, based on just need for elegance , code or law has to work first even if dirty .