What are the opening hours?

Monday 8am - 5pm Tuesday 8am - 5pm Wednesday 8am - 5pm Thursday 8am - 5pm Friday 8am - 5pm

Nuvolum

Nuvolum is an integrated marketing and PR firm that creates brand growth by developing engaging expe

Our team seamlessly works together to propel our clients to the next level of success. We analyze data to evolve our strategy and consistently elevate our expertise and help brands build experiences that win. Through video production, content writing, SEO, graphic design, and so much more, Nuvolum is the foundation of success for your brand.

11/05/2026

Read the caption. Or skip it. The caption is the trailer. The real thing is on Substack, free, link in bio. Sam Altman said it on a podcast last week. It is “quite terrible,” in his words, for the world’s limited DRAM to sit in your computer being used 5 percent of the time. The PC press roasted him. Priced In reads it differently. It is a product announcement. Three numbers. OpenAI’s stated target: 900,000 DRAM wafer starts a month. TrendForce: conventional DRAM contract prices up 90 to 95 percent in a single quarter this year. Samsung, the same day the podcast dropped: 2027 demand will far exceed supply. Put those together and the remark is not a gaffe. When the vendor calls your hardware a misallocation, your inference is going on a meter. That is the business model, said out loud. What the full edition has that this caption does not: the $1.4 trillion figure in Altman’s own words, why local inference is repriced rather than dead, and what all of it does to the invoice at a medical practice. Full edition on Substack. Free. Link in bio. Follow Nuvolum for the daily edition and subscribe on Substack. Priced In publishes every weekday. The economics of what’s next.

11/02/2026

Open this before you scroll.
The link used to pay. This week Google admitted the answer pays less. Alphabet gave back roughly $500 billion in a day. Reddit is losing a million search referrals a week and says AI Mode sends nothing back.
Here is the part that matters for your business. The response from Google, OpenAI and the publishers is the same move: put the ad inside the answer. Some ads you will see. Some are written only for the AI agent doing the reading.
Being found just moved from the link to the citation, and the citation is turning into a paid position on both ends. Who pays, who checks, and what to do about it is in the full edition.
Full edition on Substack. Free. Link in bio. Follow Nuvolum for the daily edition and subscribe to Priced In.
The economics of what’s next.

10/30/2026

Clean. Rendering the full animation now.
Open this before you scroll.
The caption is the short version. The full edition is on Substack, free, link in bio.
Google announced its strongest model on Wednesday. Gemini 4 Carbon, first on the index, $2 per million tokens. Then it closed the door. Vetted cyber defenders first. Everyone else gets it “as soon as possible,” with no date attached.
Anthropic set the pattern in April. Claude Mythos Preview goes to Project Glasswing partners only, at $25 in and $125 out. Twelve and a half times Carbon’s rate.
The entry fee is not money. It is a background check, hardware MFA, and a security team you already employ. Anthropic’s bar: a major attack on your codebase would reach 100 million people.
The seat price was never the whole bill. The second gate is who you are.
Full edition on Substack. Free. Link in bio. Follow Nuvolum for the daily edition and subscribe on Substack.
Back Monday.
The economics of what’s next.

10/29/2026

Read the caption.
Or skip it. The caption is the trailer. The real thing is on Substack, free, link in bio.
Alphabet and Microsoft reported the same night, spent about the same money on the same thing, and the market gave them opposite grades. One went $9 billion into the hole. The other cleared $18 billion.
The difference is not who spent less. It’s who builds and who rents. If your business rents AI compute from either one, the bill is yours too.
Full edition on Substack. Free. Link in bio. Follow Nuvolum for the daily edition.

10/29/2026

Open this before you scroll.
The real version is on Substack, free, link in bio. This is the short one.
Sunday night Elon Musk said more than 200 AI cloud companies, the neoclouds, face guppy-level death. The lenders are gone. CoreWeave, Nebius and IREN are each down more than 30 percent this month. Gartner says more than 80 percent of the category stops operating independently by 2030.
Then he said the line that actually moves the market. SpaceX is not buying any of them. Pointless. The one company with $40 billion in fresh GPU financing just walked away from the auction.
So who buys them?
Nvidia already answered that question thirteen months ago, in an SEC filing. A $6.3 billion order that obligates Nvidia to buy CoreWeave’s unsold capacity through 2032. Supplier, shareholder, buyer of last resort. And the same filing carries a termination clause that disappears the backstop the day a bankruptcy petition lands and sticks.
Nvidia never needed to buy the company. It bought the exit.
For every business that runs on AI, this is a vendor question before it is a market question. Your tools run on someone’s compute. Ask whose. If it is a company on Musk’s list, your contract is as stable as their lender. If it is Amazon, Microsoft or Google, expect the bill to show it.
Full edition, every filing linked, in today’s Priced In. Free on Substack, link in bio. Follow Nuvolum for the daily edition.
Priced In The economics of what’s next.

10/26/2026

Read the caption.
Or skip it. The caption is the trailer. The real thing is on Substack, free, link in bio.
Microsoft published a study in Science saying its diagnostic AI gets the hardest cases in medicine right 83% of the time. Physicians got 36%.
Here is what the headline leaves out. Every author works for Microsoft and holds Microsoft stock. No outside money funded it. Microsoft wrote the test, built the system, built the AI that graded the answers, and has a patent pending on the framework. In 2025 Microsoft said this needs governance before it touches a patient. Then it tested itself.
The FDA, HHS and ONC exist for exactly this. A drug does not reach a patient because the manufacturer liked its own trial. A diagnostic AI in an emergency department should not either.
The full breakdown, every number and every source, is in today’s Priced In from Nuvolum’s founder. Free on Substack, link in bio. Follow here and subscribe there.
Priced In The economics of what’s next.

10/21/2026

Open this one before you scroll.
A federal order signed in July. Published in October. In between, the model it bans got better and stayed free.
That is the week in AI, and it is the whole argument of Jock Putney’s Priced In in one edition. The Department of Homeland Security has excluded DeepSeek, and every product built on it, from the civilian government. Order number 2026-01, signed July 24, public October 19. Three days before it surfaced, DeepSeek shipped new weights under an MIT license at $0.30 per million tokens in and $1.20 out.
The question the edition answers is the one every business running cheap open models should be asking right now: what does a model cost once you add what it takes to be allowed to use it? Compliance review. Cyber insurance. A procurement desk that now has to prove a negative.
There is also a security firm’s look inside criminals’ dashboards, where the same model was running break-in attempts at roughly $5 each. And there is the July 30 post where Jock said this door was closing, six days after the order was signed and 81 days before anyone could read it.
Nuvolum builds strategy on exactly this kind of reading: where the price is going, not where it is. Read the full edition free on Substack, link in bio, and follow Priced In for tomorrow’s.
The economics of what’s next.

10/20/2026

Read the caption.
Or skip it and go straight to the source. Priced In is free on Substack, link in bio.
A question for anyone running a business. If the biggest software company on earth cannot negotiate a frontier AI price down, what are your odds?
Here is what the record shows. In November 2025, Microsoft, Nvidia and Anthropic signed a three-way partnership. Anthropic committed to buy $30 billion of Azure. Microsoft committed up to $5 billion of investment. Then The Information reported, citing one source, that Microsoft is now on track to pay Anthropic more than $2 billion a year to run Claude inside Copilot. Microsoft has not confirmed it, so hold that number loosely.
Here is what is not loose. Microsoft built a router to send every task to the cheapest model that can handle it. Its own partner’s models run the routine work. The hard work still goes to Claude at a premium. The bill went from a reported $500 million a year in January to a reported $2 billion by October, at the one buyer with every tool to hold it down.
That is the pricing signal for every company below Microsoft. Frontier intelligence is not getting cheaper for the work that matters. It is being metered, routed and priced like a utility, and the lab holds the meter.
The full edition covers what Anthropic’s own employment data says about who does the work now, six agent hours for every human hour, and why that widens the lab’s margin while the buyer’s bill grows.
Read it on Substack. Free. Link in bio. Follow Priced In for the daily.
The economics of what’s next.

10/19/2026

Read the caption.
Or skip it and go straight to the source. Full edition on Substack, free, link in bio.
If your business runs on AI tools, your December invoice is already written. Google’s developer page lists Gemini 4 Argon at $2 and $10 per million tokens through November 30, then $4 and $20 from December 1. Not a rumor. A published date.
The subscription ladder tells the same story from the other end. The $20 seat has not moved since February 2023. Everything above it was built in the last twenty-two months, and the ceiling now sits at $500 a month. What the expensive seats sell is not chat. It is agents, and agents are metered by the hour they run, not the question you ask.
Behind all of it, ASML says its planned 2027 and 2028 supply will not meet demand. The chips are rationed two years out, and the companies that reserved them are the companies setting your price.
Three things to do this week: budget anything that runs past November 30 at the December rate, write your AI line as seat plus meter, and read what Meta’s new open-weight model actually takes to run before you plan to self-host.
In July, Jock Putney called $1,000 a month plus token charges. The market is at $500 and the first doubling is dated.
Full edition on Substack, free, link in bio. Subscribe there, follow here.
The economics of what’s next.

10/16/2026

Read the caption.
Then read the edition. It’s free on Substack, link in bio.
Your AI tools run on someone’s servers, and those servers sit on someone’s balance sheet. This week the bond market showed which balance sheets it trusts.
CoreWeave’s bonds hit a 15.78 percent yield. Oracle sits one notch above junk. Three questions every business should ask its AI vendor are in the full edition.
Link in bio. Follow and subscribe for more.
Priced In. The economics of what’s next.

Address

435 Pacific Avenue Ste 400
San Francisco, CA
94133

Opening Hours

Monday	8am - 5pm
Tuesday	8am - 5pm
Wednesday	8am - 5pm
Thursday	8am - 5pm
Friday	8am - 5pm

Website

http://nuvolum.com/

Alerts

Be the first to know and let us send you an email when Nuvolum posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.