Google Throws the Kitchen Sink, the FDE Land Grab Spreads, & Anthropic's $900B While Customers Wait for $68B in Value
Google crams AI into everything, the FDE hiring war jumps from the labs to the platforms, and Anthropic passes OpenAI at $900B valuation
Google Throws the Kitchen Sink
At I/O on Tuesday, Google put AI into nearly everything it makes. Here are just some announcements from Sundar Pichai’s recap:
Gemini Spark, a personal AI agent in the Gemini app, with Android and Chrome coming next
Gemini 3.5 Flash, pitched as faster and cheaper than rival models, with Gemini 3.5 Pro due next month
Gemini Omni Flash, an any-input, any-output model, rolling into the Gemini app, Google Flow, and YouTube Shorts
Two new in-house chips, TPU 8t for training and TPU 8i for inference
Antigravity 2.0, reworking Google’s AI coding tool into an agent manager
A new AI Mode button in the Search box, plus agentic experiences, just-in-time UI, and the ability to build apps inside Search
AI Mode now past 1 billion monthly users, and AI Overviews over 2.5 billion
Daily Brief in Gemini, a morning digest from your inbox and calendar
Ask YouTube, which finds specific moments inside relevant videos
Ask Maps, for longer and more complex questions
Docs Live, writing a document by voice, with voice editing coming to Gmail and Keep
Google Pics, a new AI image creation and editing tool, not to be confused with Google Photos
Nano Banana for image generation, Lyria for music, Veo for video, and new editing agents in Google Flow
Gemini for Science, wired into more than 30 life-science databases, plus Deep Think and Deep Research
Audio glasses, display glasses, and XR eyewear running Gemini
SynthID watermarking across 100 billion images and videos, with OpenAI now joining the scheme
A new AI Ultra tier at $100 a month
And much more, Nathan Clark’s post summed it up well.
Google is spraying AI across every surface it owns and will see what sticks.
They already have the distribution. Gemini passed 900 million monthly users, double a year ago, thirteen Google products have more than a billion users each, and the company now runs 3.2 quadrillion tokens a month, seven times last year.
If just one of these hits, they will have a new billion user product. But how can consumers keep up with all the different products, and all the different SKUs? What are they paying for?
The good news for Google is they get paid even if these don’t land. Apple’s new Siri runs on Gemini, and Anthropic trains and serves Claude on Google Cloud TPUs under a deal worth tens of billions, which we covered two weeks ago, and Google holds roughly $10 billion of Anthropic equity.
With OpenAI focusing more on enterprise, Google has a chance to conquer the consumer market. They’re throwing the kitchen sink at this, and while all of these products and SKUs are adding up, anything that lands is upside.
The FDE Land Grab Jumps From the Labs to the Platforms
Last week we covered OpenAI’s $4 billion Deployment Company and Google Cloud’s new forward deployed engineering team. The platforms followed suit this week.
ServiceNow and Accenture launched a joint forward deployed engineering program, putting ServiceNow’s FDEs and Accenture’s industry teams inside shared customers to build agentic workflows in production before any company-wide rollout. Salesforce committed to hiring 1,000 FDEs. The title Palantir invented is becoming the most contested hire in enterprise software.
Most of us remember MIT’s study claiming 95% of enterprise AI pilots show no measurable return, and while contested for its method, the direction probably holds.
Failures come at the last mile of building and wiring into a company’s real data and workflows so it provides value. Anybody who has vibe-coded knows how easy the first 90% of a build is, and how brutal the final 10% is. Now imagine what this is like for an enterprise.
Which explains the focus on engineers who can make deployments work. Labs, cloud providers, system integrators, and software platforms are all bidding for the same small pool, which is exactly the kind of shortage that sets pay.
If the model is the easy part and the engineer is the scarce part, where do you think the money flows next?
A quick note from me.
Every company in that story is fighting over the same forward deployed engineers, so I started tracking them. FDE Pulse follows the forward deployed engineer market: who’s hiring, what they pay, and what backgrounds they’re looking for. Totally free.
Here are the top paying roles this week.
Right now there are 134 open FDE roles at a $135K median base, with the top listings from Google Cloud, ServiceNow, and the big consultancies clearing $300K. If the last mile is where AI value gets made, this is the job market making it.
Curious how this profession grows and changes? Check out the FDE Pulse →
Anthropic Hits $900B Before the AI Returns Come
Anthropic is raising more than $30 billion at a valuation above $900 billion, backers include Sequoia, Dragoneer, and Altimeter, and would surpass OpenAI’s $852 billion mark from March. Anthropic’s $44 billion run rate is up 80x in a year, while OpenAI is about $24 billion.
But what about their customers? With Anthropic’s $44 billion & OpenAI’s $24 billion ARR, how many of their customers have realized $68 billion in value this year?
Start with labor, the savings everyone points to. AI was the named reason for 49,135 US job cuts through April, per Challenger. Put a generous $150,000 of fully loaded pay on each one and that’s about $7 billion a year in payroll removed, enough to cover roughly a tenth of what those two labs alone collect.
To close the rest on headcount, companies would have to cut another 400,000 jobs and credit every one to AI. We’re at 49,000, and Gartner found the layoffs already blamed on automation are mostly failing to deliver the savings that justified them.
As for revenue, Deloitte’s 2026 survey found 20% of companies are already growing revenue from AI, with many more expecting to, but $68 billion worth?
Anthropic is the strongest case that the bet pays. Coding and back-office automation return money today, and that’s what Claude sells. If any lab’s revenue reflects value its customers capture, it’s Anthropic’s.
Step back, though, and the spending is years ahead of the payback. The money has gone in for tokens, but the returns aren’t meeting forecasts. Investors are betting that companies will keep buying Anthropic and OpenAI tokens, but when will their customers see a payback?
The Signal
3 takes that didn’t fit above, plus one bet.
1. Meta starts cutting 8,000 jobs, and we called the backlash last week The cuts began May 20, and after Coinbase...
2. Nvidia just crushed earnings and the stock fell anyway Data center nearly doubled and Jensen says demand has gone parabolic, but the reaction tells you...
3. China shipped four frontier-class coding models in a month and the West barely noticed GLM-5.1, MiniMax M2.7, Kimi K2.6, and DeepSeek V4 all landed at...
My bet: The first big AI layoff to get publicly reversed lands before...
Unlock The Signal and 3+ years of archives, $90/year →