Superstar Effects or Arm Floaties?

Something big is happening, the "taste" debate, early hiring trends of 2026, and 1,470+ fresh startup jobs

Feb 17, 2026

∙ Paid

TL;DR of this edition:

(1) Interesting reads of the week from across startups & tech, including the 80M+ view viral X article, Wispr’s CEO on job searching, and debunking the death of software.

(2) The early startup job trends of 2026.

(3) Superstar Effects vs. The Arm Floaties Theory? Over the long term, will AI increase or decrease the gap between the median employee and the top 1% employee?

(3) Relevant roles posted in the past week at startups for young candidates.

Interesting Links

Something big is happening. This article by Matt Shumer (CEO of HyperWrite) went completely viral this week, racking up 82M+ views on X. The piece was honestly fine, but nothing particularly special in my mind (even my Mom saw it and sent it in the family group chat!). The author draws a parallel between today and the calm before COVID: most people think this is overblown when we’re actually in the midst of a much larger shift.
- Matt’s key advice: pay for the best models, use them daily on real work (not just search-style prompts), automate your highest-leverage tasks, build financial resilience, and develop the habit of adapting quickly.
If you are still applying to jobs, read this. Wispr CEO Tanay Kothari (Stanford CS, Forbes 30U30) argues that candidates can skip the line by shipping work before you are hired. Builders ship features, analysts send product teardowns, operators close deals — all before submitting an application. The application is not the beginning.
- Tanay is spot on. Submitting applications into the void is a strategy doomed for failure. You need to be proactive and add value before ever applying if you're interested in working for early-stage startups. I talked about related threads a few months ago in more depth and the idea of creating an echo chamber of your awesomeness. Check it out here.
Death of Software? Nah. a16z released a newsletter drawing direct parallels to earlier tech shifts where PCs did not eliminate mainframes, e-commerce did not kill physical retail, and streaming did not end media creation. In each case, legacy systems persisted, new platforms layered on top, complexity increased, and the total amount of software, businesses, and work expanded far beyond early predictions. AI, he argues, follows the same pattern.
AI adoption splits by age and income. Morgan Stanley data shows 52% of people 55+ have never used AI for work, and 41% have never used it personally. We are still very early. See the full data below.

YC released its Spring 2026 Requests for Startups, highlighting opportunities in AI-native product management, AI-native hedge funds and agencies, stablecoin-based financial services, AI for government and fraud detection, modernized U.S. industrial systems, AI guidance for physical work, large spatial reasoning models, and new infrastructure to make training and deploying LLMs dramatically easier.
Building marketplaces in the AI era. Olivia Moore (a16z) shared more about how the fundamentals haven't changed (LTV/CAC, retention, margins still matter), but AI unlocks categories that were previously too expensive to operate. A few of her thoughts: (1) Use it as your "ops team" to handle vetting, documentation, and high-coordination tasks. (2) The best opportunities involve complex matching, not one-click purchases. (3) AI app stores are worth investing in now despite low traffic today (like mastering Google SEO early).
What’s working in AI for GTM? 40 GTM workflows where AI is driving real gains, including AI-built competitor comparison pages generating thousands of SEO landing pages, AI-powered outbound producing personalized emails at scale for under $10, and AI copilots synthesizing Gong calls and CRM data for sales prep.
The female philosopher in charge of giving Anthropic “soul.” The WSJ profiled Amanda Askell, the resident philosopher at Anthropic, who helps shape Claude’s sense of right and wrong. Her job includes studying the model’s reasoning, correcting misfires with 100+ page prompts, and guiding its personality using Anthropic’s Constitutional AI framework, a process she compares to raising a child. In response to the WSJ piece, Elon Musk, in typical fashion, commented that “Those without children lack a stake in the future” and added, “Anthropic is Misanthropic.”
Check out the last two editions of Builders Report by Teli Labs, where we spotlight standout young founding teams. In each edition, the founders themselves share insights on their business, traction, and vision for the future.
- Edition 3: Five young teams building in stablecoin payments (Crebit, a16z speedrun), AI video training datasets (Shofo, YC), skiing and snowboarding matchmaking (Shredder), real-time Gen Z social coordination (Pinned Social), and AI-powered sustainable fashion discovery (Masira).
- Edition 2: Six young teams building voice AI-powered dating (Known, $10M seed), gourmet viral slime (Sliimeyhoney, $2.9M revenue, 1.1M Tik Tok followers), AI-native video production (Cocreate, YC), real-time AI photography coaching (Serve), B2B campus security (LockGuard), and AI drug toxicity prediction (Helios).
1,470 fresh startup jobs posted in the past week. The roles specifically for young candidates are featured at the bottom of this newsletter!

Early Startup Job Trends of 2026

Teli Labs' job board tracks 600+ of the hottest VC-backed startups, with 10K+ roles currently active. Here's a pulse on the startup job market so far this year.

What is the most popular day of top startups to post jobs?

Wednesday (23% of total postings). Followed by Friday (20%). Notably, a new job posting occurred 2.6x more on Saturday than Monday.

Number of internships?

228 (1.67%). Internships remain a sliver of the market, particularly for job postings.

Which functions are heating up / cooling down?

Using January 24th as a midpoint for the year so far, we can spot some early directional shifts in hiring by function (measured as each function’s share of total postings):

Gaining ground over the past 3 weeks: Growth (+51%), Ops (+22%), Legal (+18%), Marketing (+5%)

Losing ground: Data (-15%), Sales (-13%), Engineering (-4%)

The bottom line: Nothing here signals major cracks in startup hiring. Postings across functions remain strong, though it’s worth noting that job listings don’t always translate to actual hires.

AI & Superstar Effects or Consolidation to the Mean?

One fascinating question is how AI will affect the productivity of workers across the spectrum. A key part of the viral "Something Big is Happening" is that you will fall behind drastically if you aren't on the cutting edge of using AI.

That raises a deeper question: will AI primarily supercharge the very best employees, or will it mostly buoy the weakest ones? There are two schools of thought:

Superstar Effects: The best workers can be emboldened to do so many things with AI that make them a 10-100x employee.
The Arm Floaties: AI can also bring up the rear and help the worst employees stay afloat.

My thoughts:

At the task level, everyone converges. The gap between the best and worst email, the best and worst first-draft analysis, the best and worst code snippet. These gaps shrink. AI encodes the knowledge and instincts of the best performers and makes them available to everyone else.

So the popular follow-up question becomes: what still matters? The trendy answer right now is "taste." The idea that execution is commoditized but knowing what's good, knowing which analysis to run, which design direction feels right, which product to build...that's where the alpha is.

I hate this framing. "Taste" feels unscientific and completely arbitrary. It's become a vibes-based moat where people claim "you either have it or you don't," which is exactly the kind of unfalsifiable argument that should make you skeptical. And more importantly, it's missing the point. If AI can raise the floor on execution, why can't it eventually raise the floor on taste? If taste is really just pattern recognition plus judgment about quality, that's precisely the kind of thing AI gets better at over time. The taste moat might be real today, but it's a shrinking island, not a permanent fortress.

Which brings me to the bigger question: will there really be these otherworldly returns for AI first movers vs. those who lag a little behind but ultimately pick it up once the tools are made for them? I'm somewhat skeptical.

We've seen this movie before. The computer, the internet, SaaS, smartphones. Each created a temporary scramble. Early adopters got an edge. Then, the tools got easier, built for everyday users to intuitively use, and the mainstream caught up. The gap between the median output worker and the top 1% output worker was always temporal, not structural. Nobody talks about the permanent advantage of learning Excel in 1990 vs. 1995. AI will likely play out the same way, though on an admittedly accelerated timeline.

Job RoundUp Highlights

Here are some jobs posted within the last two weeks that are a great fit for recent grads, college students, and young candidates in large US cities, including (1) Growth, Sales, Marketing & Ops, (2) Engineering, (3) Design, Data, Legal, Finance & Admin Roles.