5,000+ Tech Workers Laid Off This Week. It's Just The Beginning. | E2286
Episode
72 min
Read time
3 min
Topics
Health & Wellness, Startups, Fundraising & VC
AI-Generated Summary
Key Takeaways
- ✓On-Premise AI Pricing Model: GoAbacus prices its Go One device at $250,000 upfront CapEx supporting 2,000 concurrent users, with a Go One Max at $350,000 supporting 8,000 users. Customers who share model weights (not underlying data) receive 20% monthly discounts. Hardware refreshes annually at no extra cost, with redundant power supplies and GPUs maintaining 99.5% uptime SLA for regulated enterprise clients.
- ✓Fractional Reserve AI Training: GoAbacus runs batch training on client hardware overnight when systems are idle, collecting only model weights — not raw data — from each deployment. This distributes training costs across all clients by industry vertical. Smaller specialized language models of 3–7 billion parameters, each handling specific banking or healthcare tasks, outperform general LLMs on deterministic enterprise workflows at a fraction of the cost.
- ✓AI Adoption Drives Earnings Beats: Block cut 40% of staff in February 2026, mandated 100% AI tool adoption company-wide, and reported Q1 earnings of 85 cents per share versus 68 cents expected — a 26% beat. The company raised full-year 2026 adjusted EPS guidance 62% above 2025 levels. This establishes a measurable benchmark: AI-first restructuring can translate to earnings outperformance within one to two quarters.
- ✓Startup Formation Playbook for Laid-Off Workers: Workers laid off from AI-disrupted companies should form groups of three to six former colleagues, identify opportunities their previous employer ignored, and build three prototype products ranked by likelihood of success. Ship the first, kill it if no traction, move to the second. Stripe Atlas incorporation data shows a step-function increase in new company formation beginning Q1 2025, validating this trend empirically.
- ✓Human Identity Verification for Agentic Transactions: Variana's, a BitTensor Subnet 54 project, uses biometric capture processed cryptographically on-device to generate privacy-preserving bio-keys stored in a decentralized registry. Miners on the network are incentivized to attack detection models continuously, hardening them against synthetic identity fraud. The core use case is delegating agent permissions — allowing a verified human to authorize an AI agent to execute financial transactions with non-repudiation.
What It Covers
GoAbacus founder David Moscatelli presents on-premise AI hardware for regulated industries, Variana's tackles human identity verification for the agentic web, and the hosts analyze 5,000+ tech layoffs at Cloudflare, Coinbase, and Block — connecting AI adoption to rising earnings and a Cambrian explosion of new startup formation.
Key Questions Answered
- •On-Premise AI Pricing Model: GoAbacus prices its Go One device at $250,000 upfront CapEx supporting 2,000 concurrent users, with a Go One Max at $350,000 supporting 8,000 users. Customers who share model weights (not underlying data) receive 20% monthly discounts. Hardware refreshes annually at no extra cost, with redundant power supplies and GPUs maintaining 99.5% uptime SLA for regulated enterprise clients.
- •Fractional Reserve AI Training: GoAbacus runs batch training on client hardware overnight when systems are idle, collecting only model weights — not raw data — from each deployment. This distributes training costs across all clients by industry vertical. Smaller specialized language models of 3–7 billion parameters, each handling specific banking or healthcare tasks, outperform general LLMs on deterministic enterprise workflows at a fraction of the cost.
- •AI Adoption Drives Earnings Beats: Block cut 40% of staff in February 2026, mandated 100% AI tool adoption company-wide, and reported Q1 earnings of 85 cents per share versus 68 cents expected — a 26% beat. The company raised full-year 2026 adjusted EPS guidance 62% above 2025 levels. This establishes a measurable benchmark: AI-first restructuring can translate to earnings outperformance within one to two quarters.
- •Startup Formation Playbook for Laid-Off Workers: Workers laid off from AI-disrupted companies should form groups of three to six former colleagues, identify opportunities their previous employer ignored, and build three prototype products ranked by likelihood of success. Ship the first, kill it if no traction, move to the second. Stripe Atlas incorporation data shows a step-function increase in new company formation beginning Q1 2025, validating this trend empirically.
- •Human Identity Verification for Agentic Transactions: Variana's, a BitTensor Subnet 54 project, uses biometric capture processed cryptographically on-device to generate privacy-preserving bio-keys stored in a decentralized registry. Miners on the network are incentivized to attack detection models continuously, hardening them against synthetic identity fraud. The core use case is delegating agent permissions — allowing a verified human to authorize an AI agent to execute financial transactions with non-repudiation.
- •AI-First Mandate as Baseline Expectation: The operational standard for startups in 2026 is requiring every employee to consult AI before beginning any task and to automate any recurring workflow. Managers should build scoring systems, run them against live data feeds, and have agents generate weekly skill-update reports. Employees who cannot demonstrate this workflow within a defined timeline should be transitioned out — Block's 100% adoption mandate and subsequent earnings beat provides the business case.
Notable Moment
Block reported that engineer productivity metrics — commits and code changes per person — exceeded two and a half times baseline after mandating universal AI tool adoption across all staff. The hosts frame this as the first major public company to validate, within a single quarter, that AI-driven restructuring directly produces measurable financial outperformance.
Episode Transcript
So here's the problem. You have banks, credit unions, hospitals that wanna use AI, but they are not interested in sending their data to any of the public AI providers. They say, how do we use AI but locally within our network so none of our data leaves our infrastructure. You can make your own decision. For me, if you look at the record of some of the executives at some of the companies, it might make one wonder. And if there's only one way to really know if your information is being shared or not, which is to not share it. Twenty five, thirty days ago, and we have over 1,600 orders of this device. I wouldn't trust Biden or Kamala or Trump or JD Vance with AI regulation, for everything. Some God Kings in Washington, DC, who are on the grift, who are getting paid by these lobbyists. That system's broken. The executive branch, to your point of executive power, control over what AI models I can use Terrifies me? To no end, Jason. This Week in Startups is brought to you by I m h health. Start feeling like your best self every day. Go to imhhealth.com/twist and use the code twist to get a free welcome kit, five free travel sachets, and 10% off your order every for all your incorporation, banking, payroll, benefits, accounting, taxes, or other back office administration needs, visit every.io. And Sentry, your team should be focused on shipping features, not chasing down bugs. New users can get $240 in free credits when they go to century.io/twist and use the code twist. Alright, everybody. Welcome back to Twist for 05/08/2026. I am at my mom and dad's Brownstone. Got to see them for the full week. Got to take in four Knicks games, Atlanta, New York, New York. And tonight, I'll be in Philly. Or when you're listening to this, I will have been in Philly. Acquired courtside for two out of four. It's a really hard thing to do these days, but I am really excited for my Knicks, and I'm really excited for today's episode. Yeah. Knicks are up, two zero over the sixers. Right? Yeah. And it takes four wins. You play three series in order to get to the finals. And so we have to win 12 games to get to the finals, obviously. And we're halfway there. Six of 12, halfway, not bad. Still a long way to go. Mission not accomplished, but we really wanna get to the finals. And then we'll have to strategize from there because we have some great, incredible folks that we're gonna be competing against in the fun. Either the Spurs are okay safe. But let's get to the show. We have a lot of great guests, and we have a lot of news. So we're gonna try to whip through this docket for everybody. I know everybody's got a lot of comments. And, at the end of the show, we'll …
Get the full transcript (14,118 words) + summary by email — free
One-time email with the complete transcript and AI summary of this episode. No account needed.
One email, no spam. We’ll also show you what SignalCast does.
You just read a 3-minute summary of a 69-minute episode.
Get This Week in Startups summarized like this every Monday — plus up to 2 more podcasts, free.
Pick Your Podcasts — FreeKeep Reading
More from This Week in Startups
How AI splits startups into winners and losers | E2322
Aug 7 · 78 min
David Senra
David Heinemeier Hansson (DHH), Co-founder of 37signals
Jul 26
More from This Week in Startups
Airtable's 80% off value crash: VCs explain why it's still a win | E2321
Aug 5 · 72 min
Latent Space
Dreamer: the Personal Agent OS — David Singleton
Mar 20
Books, tools, and gear mentioned in this episode
SignalCast may earn commission on purchases via these links. As an Amazon Associate, SignalCast earns from qualifying purchases.
Tools
by Stripe
“Stripe Atlas incorporation data shows a step-function increase in new company formation beginning Q1 2025, validating this trend empirically.”
“Variana's, a BitTensor Subnet 54 project, uses biometric capture processed cryptographically on-device to generate privacy-preserving bio-keys stored in a decentralized registry.”
Gear
- Go One MaxBy guest
by GoAbacus
“GoAbacus prices its Go One device at $250,000 upfront CapEx supporting 2,000 concurrent users, with a Go One Max at $350,000 supporting 8,000 users.”
More from This Week in Startups
We summarize every new episode. Want them in your inbox?
How AI splits startups into winners and losers | E2322
Airtable's 80% off value crash: VCs explain why it's still a win | E2321
How Bespoke faked AI until it actually worked (w/ Akemi Tsunagawa) | E2320
Why AI has no taste and how to fix it (w/ Thais Castello Branco) | E2319
$100T is managed by “human duct tape” | E2308
Similar Episodes
Related episodes from other podcasts
David Senra
Jul 26
David Heinemeier Hansson (DHH), Co-founder of 37signals
Latent Space
Mar 20
Dreamer: the Personal Agent OS — David Singleton
Equity
Mar 4
How PopSockets broke the VC-backed consumer hardware mold
David Senra
Aug 9
Michael Ovitz, Co-founder of CAA
20VC (20 Minute VC)
Aug 8
20VC: The AI Boom Will Create Enormous Roadkill: Who Wins & Loses | Why Founders Should Never Take Multi-Stage Money at Seed | Why Triple, Triple, Double, Double is Good Enough
Explore Related Topics
This podcast is featured in Best Startup Podcasts (2026) — ranked and reviewed with AI summaries.
Read this week's Health & Longevity Podcast Insights — cross-podcast analysis updated weekly.
You're clearly into This Week in Startups.
Every Monday, we deliver AI summaries of the latest episodes from This Week in Startups and 192+ other podcasts. Free for one show.
Start My Monday DigestNo credit card · Unsubscribe anytime