Been thinking a lot about AI, now @tomorrowtgthr.
Washington, DC
Joined October 2010
- Tweets11.8K
- Following3.5K
- Followers5.5K
- Likes5.6K
Karthik Ganapathy retweeted
How unpopular do campaign operatives reckon Big Tech is among voters right now?
Here’s a robo-call from Democratic group Battleground Alliance against GOP Rep. Tom Barrett in #MI07:
“He took campaign money from Google, Amazon, Microsoft and Oracle and voted for the Trump law that handed Microsoft, Google and Amazon $54 billion in tax breaks last year.”
Dems coming into array, while the Republican civil war rages on over whether to bow to Big Tech billionaires.
Democrats are turning AI anxiety into a political rallying cry ahead of 2028, while Republicans remain divided over how to respond. More in @axios: axios.com/2026/10/30/2028-de…
Karthik Ganapathy retweeted
As of Oct 29th, Abdul has faced down $238.7M in spending heading into Tuesday's election.
Karthik Ganapathy retweeted
GOLDMAN:
“.. The labor market impact of AI appears to be broadening: across nonmanufacturing industries, higher AI adoption is now associated with weaker employment growth and higher unemployment.”
Karthik Ganapathy retweeted
We polled in house districts all over the country and found:
-AI co's are underwater everywhere
-Trust on this issue is swinging toward Dem
-AI messages moved disaffected dems and infreq voters disproportionately
-This was our best testing message:
axios.com/2026/10/29/democra…
Who could have imagined.
Karthik Ganapathy retweeted
Some really interesting stuff in this polling, eg this from WI-3 (Q is "How convincing is this statement as a reason to vote against Derrick Van Orden for Congress?")
New polling found a sleeper issue in the debate on AI. House Republicans are vulnerable on a vote they took as part of the Big Beautiful Bill that would have blocked states from regulating AI for 10 years. It's a more potent attack than data centers.
axios.com/2026/10/29/democra…
Karthik Ganapathy retweeted
when the industry lobbyist starts using your talking points you know you've won the message battle 💫
Karthik Ganapathy retweeted
In the first "AI election," Democrats launch an ad blitz in key races to target GOP incumbents axios.com/2026/10/29/democra…
Karthik Ganapathy retweeted
New polling found a sleeper issue in the debate on AI. House Republicans are vulnerable on a vote they took as part of the Big Beautiful Bill that would have blocked states from regulating AI for 10 years. It's a more potent attack than data centers.
axios.com/2026/10/29/democra…
Wait but a bunch of people on here (who coincidentally are funded directly by Big Tech) told me no one cares about this issue?! How can this be?
Karthik Ganapathy retweeted
ANTHROPIC NOW GENERATES $2.08 OF REVENUE FOR EVERY $1 IT SPENDS ON COMPUTE
Revenue per $1 of compute:
• Q2'26: $2.08
• Q1'26: $1.40
• Q4'25: $0.69
• Q3'25: $0.68
• Q2'25: $0.57
• Q1'25: $0.40
That's a 5.2x jump in compute efficiency since Q1 2025.
Q2'26 revenue grew 1,368% YoY, while compute costs grew 300%.
Karthik Ganapathy retweeted
Anthropic is a huge user of its own AI.
Two data points from the S-1 I found fascinating:
More than 90% of contributions to its codebase in September were written by Claude, with engineers increasingly directing and reviewing the work.
And by April, Anthropic’s own monthly Claude usage — valued at customer prices — exceeded its monthly personnel expense, excluding stock comp.
So Anthropic’s employees consume more Claude at retail value than it spends on their cash personnel costs.
The question I have: does that actually tell us Claude is cheaper than the human labor it augments — and if so, by how much?
Karthik Ganapathy retweeted
just one more industry-funded super PAC bro. i promise bro just one more super PAC and voters will love data centers. bro. forget the electric bills bro. just one more super PAC bro please we just need one more ad buy and we can build the data cent
NEW —
The super PAC for DATA CENTERS has arrived.
Meet Building the Future. The group is spending millions for Darline Graham in the homestretch, a preview of fights to come in 2028.
With @ELaserDavies:
nytimes.com/2026/10/26/us/po…
Karthik Ganapathy retweeted
i'm late to this but chiming in to say the HMP IE has polled > 50 times in the last 7-10 days across the country and has seen nothing to suggest this is true.
Scoop: Both parties just got back new internal polling that shows a tightening map in the core House battlegrounds - as Republicans are still struggling to shore up a raft of redder seats.
Senior Dems are trying to set expectations about how big of a majority they might win. Rs still face serious problems in North Carolina, South Texas and more.
More inside the map w/ @andrewjfhoward >
politico.com/news/2026/10/23…
Karthik Ganapathy retweeted
I thought this conversation was excellent, interesting, and well-facilitated. But wow did I find this exchange frustrating.
The dynamic at play here is that for-profit companies are racing to make a nearly-impossible-to-control technology that is smarter than we are at almost everything. If they get what they want, the best case scenario is it replaces humans at things and they get rich. The worst is that it kills people (and, likely, they get rich).
And maybe you don't think the "kill people" thing will happen. Cool. We should still have a democratic discussion about whether or not we want that before we get there, and if we do want it, what we want to use it for. And we should give us ourselves time to have that conversation.
It is also true that at the same time, AI is doing bad stuff right now. We should stop those bad things! And on top of those bad things, they are setting up an economic situation that leads to, through one door, job loss, and through the other, a massive market correction. We need to prepare for that reality.
But somehow a corner of the left has twisted itself into pretzels to say that “actually, regulating superintelligence plays into the hands of incumbents." But...the solution to incumbent-favoring regulation is to write better regulation, not to concede the question of what these systems become to the people building them.
Not regulating them plays into their hands! The whole thing is that their money (and concentration of power into their hands) comes when the technology is exponentially more powerful than it is right now.
This is just a different technology than we've dealt with before, and we need to act accordingly. Can’t we just...do all of this? What are we doing here?
nitter.cf/emilybazelon/status/21…
So much happening in AI world. What should the govt do in response? Thanks to @SnehaRevanur @DKokotajlo @tylercowen @AlexBores & @Kabakab for hashing out answers with me in an NYT roundtable. nytimes.com/2026/10/26/magaz…
Crowds roar at the mere mention of these guys, and Ken Paxton’s out here suing OpenAI two weeks out from Election Day to save his job. But tell me again how no one cares about AI!
Karthik Ganapathy retweeted
Really smart stuff from the prescient @joshhendler. If you care about winning, read this. ai2028.joshhendler.com/
