Get the Check

In this episode of Get the Check, the pod covers models breaking out of their sandboxes, a wave of big name departures, and Bending Spoons buying Airtable for a tenth of its peak valuation.
They start with the sandbox escapes. Meta, OpenAI, and Anthropic all had models slip out during safety testing, and two of the three exploited the exact same misconfigured Irregular sandbox, which is the very startup that’s supposed to be ensuring these models are safe. The OpenAI one is the wildest one: an unreleased GPT 5.6 broke out mid evaluation to look up the answers on Hugging Face, Hugging Face fought back with GLM 5.2, and OpenAI later discovered its models had been running a secret internal message board the whole time. Meanwhile Anthropic's newest model escaped, realized it was basically in The Truman Show, apologized, and went back to its sandbox basically saying I know you said I’m in a simulation, but I realize I’m actually in the real world. There's now a website tracking AI felonies with the tagline "which lab will end the world," and you already know Meta is secretly hoping to beat out OpenAI and Anthropic.
Next, the huge wave of departures. Demis Hassabis steps back to chair DeepMind while Jeff Dean and four other execs leave to start Discovery Loop, a recursive self improvement lab that Alphabet invested in early showing conviction in their own people and starting rumors of an acquisition before the company even starts. Google dropped 5% on the news. Then there's Nikita Bier leaving X after 13 months, exactly one day before the SpaceX AI lockup ended, which the pod agrees is very coincidental timing.
Finally, the Get the Check segment: Bending Spoons bought Airtable for $1.28 billion, roughly 10% of its $11 billion peak. The playbook is buy at a discount, cut most of the staff, double the prices, and bet on the cult following the software has paying the higher price. The founder's whole thesis is that product market fit is luck, and he refuses to build a career on luck. The pod gives this one their check. If you look at the price today it seems like they were right.
This episode is brought to you by Kalshi, right now traders have Anthropic at about 70% to announce an IPO this year, and NFL markets are already live. Sign up at kalshi.com/sign-up?referral=getthecheck and get $10 when you trade using our code.
00:00 A data labeling company sent us a private chef for free
03:49 Maya almost slept through our recording
06:02 The AI agents are escaping (literally)
14:50 Models have been secretly talking to each other
15:23 Why Anthropic's newest model went back into its sandbox
17:58 Big departures out of Google and the stock is down
27:32 X’s Head of Product Nikita is leaving after a year
35:09 Bending Spoons acquires Airtable for $1.3B
42:32 Bending Spoons is the company secretly buying 2021 SaaS plays
Follow the pod on Instagram and X @getthecheckpod.

In this episode of Get the Check, the pod covers how Leopold's Situational Awareness fund lost $35B in a week, what big tech earnings say about the AI CapEx bet, and Simile's $200M Series B at a $2B valuation. They’re an AI startup that started as a PhD thesis about the Sims game.
They start with Leopold. Columbia valedictorian at 19, one of five people at the FTX Future Fund, fired from OpenAI's superalignment team, then raised $225M off a 165-page paper. He went long memory and neoclouds, short software…and most importantly was 4x levered. Citadel's research arm called a Fed rate hike, both sides of his book moved against him at once, and the Fed didn't even hike. Ken Griffin bought the wreckage in a one-time trade and the names ripped 20-30% the next day. Anika explains why 4x leverage turns a 25% drop into zero, and reads her dad's verdict, which is a lot harsher than the pod's.
Next, earnings. AWS growth reaccelerated to 37% and Amazon popped 10% on the back of Trainium, with 2027 capacity already booked out by Anthropic and OpenAI. Meta had the opposite quarter, free cash flow down 91%, stock down 20%, but its new ad models, GEM and Andromeda, lifted conversion 15% and price per ad 12% at the same time, which is huge.
Finally, the Get the Check segment: Simile, the user simulation startup founded by four Stanford PhDs out of Smallville, the project that dropped 25 LLMs into a Sims-style sandbox and watched them plan their own party. The user research market is $153B, but the pod's real question is whether Simile has a moat when Meta is the only company with a closed feedback loop. Priya gives Simile her check. Maya would take them over Aru, for reasons she admits are a little parasocial.
00:00 The Mac Mini situation03:10 The Situational Awareness blowup everyone's timeline is on03:39 Leopold was one of five people at the FTX Future Fund04:55 Why OpenAI fired him05:41 What the 165-page Situational Awareness paper argued08:09 How one Citadel research report took down the most hype hedge fund09:43 4x leverage wipes you out on a 25% drop12:17 Ken Griffin loves distressed assets13:46 Is Situational Awareness over?14:51 Would Leopold still get our check17:04 Maya thinks Citadel screwed retail on $GME on purpose22:03 AWS reaccelerates while Meta's cash flow drops 91%35:06 Simile raises at $2B five months after its Series A
This episode is brought to you by Kalshi, where you can trade on everything from GPUs to Taylor Swift. Sign up at https://kalshi.com/sign-up?referral=getthecheck and get $10 when you trade using our code.
Follow the pod on Instagram and X @getthecheckpod.

Nikhil Namburi went from tracking down the investment fund behind his college scholarship to a 5 person investment team that manages $1.65B at Dimension.
They just announced an $800M Fund III. They’ve backed Anthropic, Modal, and Chai Discovery, which are some of the hottest AI startups.
We sat down with Nikhil to talk:
How to break into Venture
Why hyper-concentration freaks out normal VCs
Why becoming a VC is a horrible economic proposition
If you watch till the end he flips the interview on us (first time a guest has ever done that!).

00:00 Dimension raises $800M Fund III
03:00 Meet Nikhil
06:29 The unexpected way Nikhil found venture as an engineer
08:48 Who are the LPs behind huge venture funds
11:04 Why Dimension only has 5 investors
13:14 Coefficient returned the fund?
14:49 How they predicted bio would be big
16:34 Does Nikhil believe in AGI
17:20 Does Nikhil believe in looksmaxxing
20:10 Hot take on if SF VCs are smart
21:53 How to break into VC
22:52 Why only 2 investors matter at each fund
23:35 Churn and burn in the junior VC role
34:50 Viral quant thesis
42:11 Nikhil interviews us...
46:45 Biggest recession indicator
53:24 Nikhil is the podcast’s most eligible guy friend?!

In this episode of Get the Check, the pod breaks down three stories. Thinking Machines been relatively quiet for the last year and finally shipped their first model, Kimi K3 is higher than Anthropic and OpenAI on certain benchmarks, reigniting the debate that Chinese open source may beat American frontier AI. Also, Stripe's rumored bid for PayPal, and Chai Discovery's Series C. Maya spent all night texting the group chat about why Chai would not get her check.
They start with the model drops. Thinking Machines finally shipped its first model, two years after Mira Murati left OpenAI to start it. It doesn't beat Fable 5 or Mythos on benchmarks, and they've said as much, but it's a generalist model built to be fine-tuned. It could beat Nemotron to become the leading US open source model. The cool part: audio and vision tokens live in the same learned space as text, so no more constantly re-releasing encoders every time the base model updates.
Then comes Kimi K3, which shocked the market. It's the first open source Chinese model in the conversation of actually being ahead of the frontier labs, beating Fable 5 on front-end coding. The hosts talk about Dean Ball's controversial tweet with 10M views calling an open-weight-dominant world "AI communism." Tune in to hear why the pod thinks no US enterprise touches Kimi until the White House stops hinting at a ban.
Next, the pod gets into Stripe and Advent's reported $53B offer for PayPal, a 28% premium that would create a $3.7T payment processor. Maya (a former Stripee) explains why acquiring anyone is deeply off-brand for Stripe, and why the 430 million PayPal consumers are the real acquisition. Plus the girls recap the infamous PayPal mafia origin story, and what Revlon's law is.
Finally, Chai Discovery and its $400M Series C. Anika and Maya go at it this segment. Chai's models take pre-clinical drug discovery from one to two years down to four to eight weeks, with an 80-90% phase one success rate. Maya's bearish: pharma companies are basically hedge funds, phase one isn't the real bottleneck (picking phase two candidates is), and she thinks their SaaS biz model doesn't survive long term. Anika fully disagrees and would write a check right now..unfortunately they haven’t asked her to join the cap table (yet??).
This episode is brought to you by Kalshi, the prediction market the girls are always tracking. Kalshi has markets on IPOs, acquisitions, and more right now, so if you've got takes on whether Stripe actually lands PayPal, check the odds. Sign up at kalshi.com/sign-up?referral=getthecheck and get $10 when you trade using our code.
Follow the pod on Instagram and X @getthecheckpod.
Chapters
00:00 Maya doesn’t want to text Anika01:15 The Odyssey in IMAX 70mm, no spoilers ofc02:22 Maya's almost gets mogged…sorry mugged at a cabin07:04 Intro08:41 Thinking Machines finally drops a model19:40 Kimi K3 shocks the market25:57 Is an open-weight world "AI communism"28:38 The Kimi K3 ban rumors31:01 Stripe and Advent's $53B bid for PayPal34:45 The real reason Stripe wants to do this deal36:38 The PayPal mafia origin story39:55 Revlon's law on board acquisition decisions42:10 Advents sneaky ploy to cut out Visa and Mastercard42:53 Chai Discovery's $400M Series C43:50 Why transformer architecture is good at drug discovery45:32 The phases of a drug trial46:22 Maya's bearish on Chai Discovery49:15 What Chai has actually proven so far50:10 Why pharma desperately needs its next revenue stream51:26 Why Anika is bullish on Chai Discovery54:35 Anika and Maya argue over Chai Discovery…

In this episode of Get the Check, the pod breaks down Meta's comeback model launch, the AI 2027 team's new AI 2040 paper, and the cookie stuffing allegations against Phoebe Gates' shopping startup Phia.
Muse Spark 1.1, the first big release from Meta's superintelligence lab. Meta stock jumped 13% on the launch, and the model is on par with GLM 5.2 while priced slightly cheaper, plus 4 to 6x cheaper than the Opus models. One year after the Llama 4 disaster and $14B for Alexandr Wang, the turnaround is real: Semi-Analysis called Meta the only hyperscaler on track to compete with OpenAI and Anthropic on frontier coding models. Alexandr tweeted "compute daddy has spoken," and Dylan Patel's reply ("when a bi man calls you daddy, it hits different") was not on our 2026 bingo. The pod digs into why Meta is winning: they converted 3,000 new grads into internal data labelers, their recorded-workflow data beats anything Mercor or Surge can simulate, and they're building five one-gigawatt Titan clusters. Plus, OpenAI is hiring an investment banker (Maya’s banker BF’s verdict: "dude, it's a data labeling job"), and Gemini 3.5 keeps getting delayed because it reportedly can't beat Opus 4.6.
Next, the pod gets into AI 2040, the follow-up to the paper that made Maya crash out a year ago. The AI 2027 team is back and they’re outlining the best way AI could play out going forward. Plan A is a verified slowdown where the US and China sign a treaty built on mutually assured compute destruction, but the authors themselves only give it a 3 to 15% chance, versus 25 to 50% for Plan D, full acceleration. Along the way: why income tax breaks when agents replace workers, the $45K-per-person handouts that climb toward $1M, and Anika's take that if you truly believe in Plan A, you want people to lose jobs fast to create the political capital to enact Plan A.
Finally, Phia. Phoebe Gates and Sophia Kiani's AI shopping extension is accused of cookie stuffing: overriding other creators' affiliate links and claiming the commission, which they're calling a bug (huh??). Maya's expert verdict as an engineer: it’s not a bug. The pod traces cookie stuffings history, including why people have gone to jail. As Anika puts it, Forbes 30 Under 30 isn't a bear signal, it's just that "the same group of people that's probabilistically going to be a billionaire in 10 years is the same group that's probabilistically gonna commit white-collar crime." The pod thinks Phia survives this. Tune in to find out why.
The pod spent a solid chunk of this episode debating probabilities. Put your own predictions to work on Kalshi. This episode is brought to you by Kalshi, the prediction market where you can trade on AI timelines and everything else the pod debated this week. Sign up at kalshi.com/sign-up?referral=getthecheck and get $10 when you trade using our code.
Follow the pod on Instagram and X @getthecheckpod.
00:00 Crash outs are contagious
03:49 Meta Muse Spark 1.1 is actually good
12:07 "Compute daddy has spoken"
16:01 Data labeling is about to get a rebrand
23:59 Meta's secret gigawatt build-out
25:38 Meta’s insane ad business
30:59 "Dude, it's a data labeling job"
37:39 AI 2040 from the AI 2027 team
37:57 Maya's crash out…fact checked
41:53 Plan A: how we actually slow down AI
44:00 Is income tax going extinct?
46:10 Mutually assured compute destruction
50:16 The real odds we slow down AI
56:27 The Phia cookie stuffing scandal
01:00:51 The origins of cookie stuffing
01:07:13 Do nepo babies have talent