Weng

Companies are hiring like crazy for Generative AI talent. In this 🧵 I’ll dive into: - Companies leading the hiring surge (+1 that stands out) - Technologies/cloud services most mentioned in these jobs - How much $$$ companies are paying for GenAI talent - Common use cases

@aschmelyun Yeah it reminds me of developers who love to spend a ton of time ensuring their codebase is 100% elegant, everything scales, and is automated, etc. That’s great… but you could’ve used that extra time in actually shipping? Marketing? Sales?

Feel like I'm missing out because I don't use skills, or a lot of MCP, or multi-agent orchestrations when using AI dev tools. I'm just like "implement this feature" or "how do this work" or "no not like that, do this instead". Idk, I feel fast and accurate so why change?

Meanwhile, the # of websites that are using $NET as a CDN has also surged, as the need to protect their data/content from AI agents and bots keeps rising. The # of new websites using Cloudflare in the past 3 months is up 57% from the same period a year ago

$NET continues to be one of the biggest beneficiaries of AI. The # of new Zero Trust instances (CloudFlare Access) in the past 3 months is up 86% YoY as companies are scrambling to protect their AI agents, MCP servers, and applications (Source: Bloomberry)

@burkeholland @k_sasha Thanks for the clarification! I'll update the chart since what I tweeted was inaccurate. I'll also delete it! Thanks again for bringing this up to me!

@k_sasha @AznWeng x.com/pierceboggan/s…

@Grauwacht Cursor isn't available in VS Code, so hard to really measure it against those other 3

@AznWeng Where's Cursor? Hovering out of sight?

@Julien_Bouvier This is measuring just VS Code extension installs (a subset of all installs, ie. CLI, etc)

@AznWeng I don't get the chart title. Are you measuring just vscode *extensions* or are you comparing full stand alone apps (like Claude Code terminal)?

@Temp002a Right, no metric is perfect - I don't think that exists. but this is close to apples to apples as we're going to get.

@AznWeng "in VSCode" is pretty limited metric. Copilot now has CLI that is seperate from VSCode, and the other two are not VSCode native.

@SAlexashenko Yep, I would say so!

@AznWeng 3rd? they have to be behind cursor right?

@k_sasha Hi Sasha, here's the source: bloomberry.com/coding-tools.h… - this is basically tacking VS Code installs over time (straight from the VS Code marketplace)

@AznWeng I am surprised noone is asking for (detailed) source of the data. The top domain listed on the infographic is not it. Are you able to share the data source?

@Jakewk Thanks! Do you have any idea on much that would impact installs of Copilot? Do you need Copilot installed to use it?

@AznWeng The dropoff actually reflects the shift from CoPilot to AgentHQ. You may want to update your graph to include the AgentHQ adoption going forward.

For those interested in tracking this daily. Here's the dashboard: bloomberry.com/coding-tools.h…

@econcallum Unless they give us specific definitions of what 'using AI' means (to us, and to the ppl they give the survey to), this data is meaningless.

Share of US businesses using AI is still stagnating

My deep dive article on $AMZN and why it is my biggest position: 1. Value of e-commerce in agentic shopping 2. Disruption of the Ad business 3. Agentic AI lead: MCP server adoption showing a strong lead for AWS vs peers 4. AWS (growth of traditional and AI compute). 5. Importance of Trainium 6. Valuation breakdown and the discount the market is giving uncoveralpha.com/p/amazons-valu…

@ysu_ChatData @bibryam (I compiled the data) - I wrote up on some findings for auth patterns here: bloomberry.com/blog/we-analyz… I didn't really analyze whether any were thin wrappers, but the average # of tools for all of them were just 5 tools, so it seems most are still in the experimentation phase

@bibryam Huge jump in server count. I am curious what you learned about duplication: how many are thin wrappers around the same upstream, and how many are actually maintained. Also, what did you see for auth patterns and rate limiting defaults?

Analyzing 1400 MCP servers – here’s what I learned 425 → 1,412 servers in 6 months (+232%) bloomberry.com/blog/we-analyz…

@tonybuildsai @TukiFromKL That is well on its way to happening

@TukiFromKL zapier is nervous about the wrong thing. their google integrations are maybe 5% of what they do. the real disruption is when every SaaS ships their own MCP endpoint.

Do you understand what Google just did? > They released a CLI that gives AI agents direct access to your Gmail, your Calendar, your Google Drive, your Sheets and your Docs > This means an AI agent can now: Read your emails. Schedule your meetings. Organize your files. Edit your spreadsheets. Draft your docs. > Every "workflow automation" SaaS charging you $49/month just became a free npm install. Zapier is shaking. 💀

@luishXYZ @signulll Its definitely biased towards tech companies #bullet5" target="_blank" rel="nofollow noopener">bloomberry.com/blog/the-state…

@signulll To be fair: Ramp is disproportionately used by tech startups and scaleups. So their invoice data might be skewed towards early adopters and not paint the full picture.

this is quite a ridiculous chart if accurate.

@HenryLSchuck @ZoomInfo @claudeai What's wild is that ZoomInfo is ranking above Clay for "GTM Engineer" even though the latter coined the term.

Today, @ZoomInfo's partnership with @claudeai has reached a new level. 45 seconds is all it took me to fully enrich a list of companies... without leaving Claude. I dropped in a CSV with company names and websites. Asked it to pull headcount, revenue, HQ location, last funding date, and total funding raised. ZoomInfo's MCP server did the rest and I downloaded a fully enriched Excel file in less than a minute. Here's why I think this is a big deal: ZoomInfo's data no longer lives just inside ZoomInfo. It follows you. Into Claude. Into ChatGPT. Into Gemini. Into the vibe-coded apps your team builds internally. Into your CRM. Into your marketing automation. We've spent 19 years building the world's best B2B data and now it can go wherever you work.

👀 for granular, technical analysts... or VCs mapping early trends: @AznWeng on MCP adoption - Early but *rapid* ramp - Read throughs for hyperscalers esp $AMZN $MSFT + $NET Some excerpts here, full study here: bloomberry.com/blog/we-analyz…

For the 1st time, Anthropic has *more* engineering jobs than OpenAI. The # of engineering jobs in Anthropic has tripled from the same period last yr (~flat for OpenAI). At the same time, OpenAI's engineering mix is shrinking (28% of jobs) while Anthropic's is growing (38%)

@jbobbink Is there a benchmark to compare the organic visibility of sites that use AI tools vs those that don't? The entire industry overall has seen less organic visibility in the past year.

I keep hearing SEOs say "but our AI content is ranking fine." That's exactly the problem. Tomek Rudzki from Peec AI just published research that should make every content team pause. They analyzed companies using popular AI content generation tools and found that 36% of the brands in their success stories had massive Google visibility drops. For one tool, it was even worse. 75% of the showcase clients had significant traffic losses. The pattern is always the same. Rank, bank and tank. Publish hundreds of AI articles. Rank quickly. Monetize. Then get hit by a core update or manual action. Abandon domain. Repeat. Google's March 2024 update deindexed 837 sites overnight. Originality AI confirmed 100% of affected pages had AI-generated content. 20 million monthly visits gone. But here's the part most people miss. When Grokipedia lost its Google rankings in early 2026, Peec AI's Malte Landwehr tracked what happened across AI search engines. ChatGPT, AI Mode, and AI Overviews all reduced citations at the exact same time. One Google penalty now makes you invisible everywhere. LLMs use search engines for grounding. If Google demotes you, ChatGPT stops citing you too. Now let me be honest here. I work in industries like casino where rank, bank and tank is a common and accepted tactic. You build a site, extract value, and move on when it gets hit. I'm not going to pretend that doesn't work. In some niches it's simply part of the business model. But here's what bothers me. The companies selling AI content generation tools are marketing them to brands building long-term businesses. SaaS companies. Ecommerce brands. B2B companies. And they're not being transparent about what happens next. Rudzki found that 3 out of 4 major global brands featured on one tool's website had suffered significant visibility losses. These aren't disposable affiliate sites. These are real companies with real reputations. After 30 years in SEO, I see a clear line: Rank, bank and tank as a deliberate strategy with eyes wide open? Fine. That's a business decision. Selling AI content at scale to brands without warning them about the risk? That's irresponsible. The data backs this up. NP Digital tested 744 articles across 68 websites and found human-written content generates 5.44x more traffic over time. Ahrefs analyzed 600,000 pages across 100,000 keywords and found only 4.6% of top-ranking pages are purely AI-generated. And Lily Ray predicts a huge crackdown on scaled AI content in 2026. If you're a tool provider or consultant, be honest with your clients. Tell them what the research actually shows. Let them make an informed decision. Rudzki's advice is simple: before you publish, ask yourself. Would I want to read this? Does it add something you can't get from ChatGPT itself? If the answer is no, don't hit publish. Sources: → Tomek Rudzki, "The real risk of AI-generated content" (Peec AI, Feb 25 2026): peec.ai/blog/the-real-… → Originality AI, AI content penalty analysis (2024): → Neil Patel / NP Digital, "AI vs Human" study, 744 articles, 68 sites (2024) → Si Quan Ong & Xibeijia Guan / Ahrefs, 600K page analysis (Jul 2025)