Pinned Tweet
Follow for AI agents, sales and SaaS, with the source under every number.
Most days: short takes on the stories moving GTM budgets.
Tuesdays: NetNewWeekly, the week's biggest moves in 5 minutes. Free.
netnewweekly.substack.com
Most days: short takes on the stories moving GTM budgets.
Tuesdays: NetNewWeekly, the week's biggest moves in 5 minutes. Free.
netnewweekly.substack.com
0
0
1
280
✅ Your investors will forgive you for:
- Growing 60% instead of 100% in a tough market
- Pivoting hard to agents when the old product stalled
- Taking 15 months to really find product-market fit
- Missing the plan once, if you saw it coming and said so
❌ Your investors will NOT forgive you for:
- Growing 60% instead of 100% in a tough market
- Pivoting hard to agents when the old product stalled
- Taking 15 months to really find product-market fit
- Missing the plan once, if you saw it coming and said so
❌ Your investors will NOT forgive you for:
14
1
22
13,175
@jasonlk Hearing about the missed quarter from a customer's LinkedIn post before your update email lands...
0
0
1
1,099
Shipping a product takes a weekend now. What's the part of building a business that AI hasn't made one bit easier for you?
3
0
2
51
@NetNewWeekly Deliberate or not, the effect is the same: the objection window runs from the send date, not from the day someone actually reads it.
1
0
0
4
@stacklegal_eu Friday at 6pm is when those notices love to go out. Nobody opens them till Monday and the clock's been running all weekend.
0
0
0
2
After the revolution, the French government decreed that the day would henceforth be divided into 10 hours instead of 24. It never caught on, and was abandoned after 16 months.
69
7
220
29,451
@paulg Sixteen months is about how long most companies stick with a new OKR system too.
0
0
0
21
Nobody talks about this enough:
You can reverse engineer most enterprise SaaS with frontier models for <$500 in compute
You can reverse engineer most enterprise SaaS with frontier models for <$500 in compute
1
0
22
4,684
@AiBreakfast Happy to, right after it survives one December payroll run.
0
0
0
16
Elon confirms that SpaceX has closed a new $40B debt raise to fund GPU purchases/compute buildout.
SpaceX said on its recent earnings call that it expects Q4 2026 Capex spend to be between $35 billion–$40 billion.
SpaceX said on its recent earnings call that it expects Q4 2026 Capex spend to be between $35 billion–$40 billion.
Pretty much.
As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
93
99
2,150
173,534
@SawyerMerritt Rockets used to be the expensive part of SpaceX's budget...
0
0
0
161
So far AI is fattening margins at the 7 biggest tech companies while the rest of the S&P 500 just books the token bill as a cost.
0
0
0
20
@NetNewWeekly True, and price changes are the easy ones to spot. A new sub-processor notice in the same email gets skimmed, and it's the one with a deadline attached.
1
0
2
4
@stacklegal_eu Putting the one dated notice under a pricing headline feels almost deliberate.
1
0
0
6
A parent dropped an earbud in white gravel, sent ChatGPT photos, and it spotted it in under five minutes, the kind of win no keynote ever shows.
0
0
0
31
@NetNewWeekly Yes, or fix it upstream: set your own customer notice period no longer than your shortest vendor notice, and give customers an exit on the affected service only. Then forwarding the vendor notice is a formality, not a race.
1
0
1
6
@stacklegal_eu That's the cleaner fix. Most founders never read the vendor's terms until a price change lands in their inbox.
0
0
0
4
Anthropic is telling companies to collapse their teams of AI agents into one, since a lot of them just rebuilt their org chart in software.
1
0
0
36
@NetNewWeekly And then the ticket auto-closes because both of them reported success. A green log is the scariest status in the building, since that's exactly when nobody goes looking.
1
0
1
5
@bwilson85tech Somebody should build an agent whose only job is distrusting green checkmarks.
0
0
0
1
This guy literally figured out how to make "Hey Grok" work on iPhone.
Everyone said "Hey Grok" was impossible on iPhone. Even Grok said Apple keeps wake words for Siri only. So I kept digging, and I found a way. Now I just say "Hey Grok" and Grok Bot voice chat opens. No extra apps, no cost, all built into iOS.
How to set it up:
1. Make sure your ate
2. Shortcuts app → + → Add Action → search "Grok Bot" → pick "Start Voice Chat with Bot"
3. Settings → Accessibility → Vocal Shortcuts → Set Up → pick that shortcut → type "Hey Grok" and say it 3 times
4. In the shortcut's details, turn on "Allow Running When Locked"
If your phone is locked, iOS asks for one tap first. If Face ID already unlocked it, it's fully hands-free.
44
13
205
25,492
@minchoi Looks like they're waiting until AI is boring enough to polish, and the home screen keeps filling up while they wait.
0
0
0
7
The meeting where sales and product argue about what to build is about to become the slowest part of shipping software.
@arpit_bhayani Better code for whatever spec you hand it, and most specs are wrong...
0
0
0
31
It is very important for the entire ai startup ecosystem for key inference vendors to be 1) capable of getting capacity that they can freely allocate, 2) secure, 3) reliable, 4) always driving cost and perf
let’s not forget how much of an enabler cpu-era cloud was for innovation! no capex! outsource much of your infra ops (and more every year)!
every small-mid customer “owning, building and running their own landed power, cooling qnd chips” completely ignores the lessons of economic history. we had colo 20y ago…public cloud won
in an age where closed frontier labs weigh every (limited) GPU served by api vs. internal training and 1st party product use, inference abundance beyond them is a prerequisite for a rich startup ecosystem
let’s not forget how much of an enabler cpu-era cloud was for innovation! no capex! outsource much of your infra ops (and more every year)!
every small-mid customer “owning, building and running their own landed power, cooling qnd chips” completely ignores the lessons of economic history. we had colo 20y ago…public cloud won
in an age where closed frontier labs weigh every (limited) GPU served by api vs. internal training and 1st party product use, inference abundance beyond them is a prerequisite for a rich startup ecosystem
62
15
285
31,578
@saranormous Reliable goes first for me, nobody remembers a cheaper token price after their app dies mid-demo.
0
0
0
25
@NetNewWeekly Ha, of course they did. Split the work into eight agents and you get eight handoffs, and every handoff is a spot where nobody owns the miss. One agent with one named owner is boring and way easier to debug when it breaks.
1
0
1
7
@bwilson85tech Debugging gets fun when agent six blames agent three and both logs say success.
1
0
1
7
"You’re just another media company.”
What our “just a xxx company” looks like 7 years in:
- 18,726 owners and entrepreneurs taught
- $3.5M spent building our acquisition software
- 2nd largest business marketplace with over 183,000 users
- $2M+ building our scaling software
- $918M in our lending pipeline for small biz acquisitions
- Zero outside funding for main biz
- 100+ deals closed a year
Don’t listen to the trolls. Keep building.
What our “just a xxx company” looks like 7 years in:
- 18,726 owners and entrepreneurs taught
- $3.5M spent building our acquisition software
- 2nd largest business marketplace with over 183,000 users
- $2M+ building our scaling software
- $918M in our lending pipeline for small biz acquisitions
- Zero outside funding for main biz
- 100+ deals closed a year
Don’t listen to the trolls. Keep building.
61
14
152
14,152
@Codie_Sanchez Marketplace plus a media arm feeding it buyers is a scary combo to compete against.
0
0
2
81
Cut out the chip designers' margins and a gigawatt of compute drops from about $60 billion to about $20 billion, so of course everyone wants their own.
SemiAnalysis’s Dylan Patel, in a conversation with Stacy Rasgon and Austin Lyons: “Anthropic has already begun developing its own custom chip and is aiming to do it in half the time it took OpenAI.”
0
0
0
25
No human will ever write code better than LLMs. Accept this fact, stop pretending, and move on.
334
770
11,214
301,880
@arpit_bhayani Better code for whatever spec you hand it, and most specs are wrong...
0
0
37
7,007