hi bud buds,
after my recent op-ed praising ai coding startup cursor, a few people asked if i got paid to write it. i wish… i want that sweet, sweet ai money. the truth is simpler: i’m just a huge fan, and i think everyone in tech should be studying them closely:
did my op-ed help them close the round? probably.
here’s my take on the fundraise:
the good:
cursor’s growth numbers in their funding announcements are bloody mental mate. revenue jumped from ~$200m to $1b in under six months. they have millions of users and 50k+ teams using, including most of the fortune 500.
amid all the naysaying about an ai valuation bubble, cursor’s unprecedented adoption and revenue growth are undeniable proof of real end-user value being delivered by ai.
the bad:
at first glance, $30b feels like a steep valuation for what is essentially a thin ai wrapper around a generic ide. however, the comps look okay. gitlab (a publicly traded dev-tools company) trades around 7x ltm sales. for cursor to grow into its valuation, they’d need to roughly 5x revenue. honestly, not crazy; they literally just did a 5x in the last six months.
my worry is cursor’s competitive position. they don’t possess pricing power, as they face competition from all directions: cheaper upstarts, internal tools, open source, and massive incumbents like microsoft/vscode with existing distribution. as inference costs fall, cursor (and their competitors’) prices will fall too. and when prices fall, revenue growth can stall fast.
the ugly:
as part of the funding news, cursor announced they’re building their own foundation model, composer. i get why: they seek differentiation for the exact reasons i mentioned above. however, i absolutely hate the strategy.
model progress is slowing while costs are exploding. openai and anthropic each burn billions of dollars annually for diminishing marginal returns. gpt 5 felt like a flop. the idea that cursor will spend a huge chunk of their $2.3b fundraise building their own model is disappointing. what happens when that model is only slightly better than existing options? will the spend have been worth it? i worry the answer is no.
tldr?
i helped cursor raise a fat round. they’re growing at a truly insane rate, but lack of competitive differentiation threatens their future growth. their answer is to build a foundation model, but i could not disagree more.
have a great week, y’all. i love you.
— dj
submit a release for the newpixels newsletter here.
big news! huge!
openai is testing chatgpt group chats, allowing up to 20 people in invite-only conversations
google released agentic checkout, letting users set a budget price for items and automatically purchase them once they fall within budget (final confirmation still required)
apple is beta testing digital id, letting iphone and apple watch users carry a digital copy of their u.s. passport in apple wallet for tsa checkpoints across 250+ airports
google released inspirational images tab in their mobile app, offering personalized image recommendations. feels like an early step toward competing with pinterest (!!)
personal faves
amazon music is testing fan groups in canada, letting users join groups (by genre or artist) to chat, post, and share songs. music is super social, so i’m excited about this
cash app released moneybot, an ai assistant that answers finance questions about your spending, income, and savings. big fan of ai helping people understand their finances, which can be scary!
spotify released audiobook recaps, letting users quickly catch up on a story to the point where they stopped listening. great use case of ai
openai released reduced em dashes in chatgpt to make ai-generated text look… less ai-generated. long overdue!
linkedin released ai-powered people search, enabling natural-language queries to find the right folks (e.g. “who are graphic designers in my network that can help me with my pitch deck?”). another great use case for ai (extremely fuzzy search)
airbnb is testing kitchen stocking service, letting guests order groceries via instacart before or during their stay, directly from the airbnb app. nice partnership 🙂
nice job guys
sofi released crypto trading inside their apps
fanduel announced prediction markets, launching in december
tiktok announced bulletin board, a new way for brands and creators to share public messages with followers
google released quarterly pixel phone updates, including ai notification summaries, power-saving maps mode, ai photo edits, and expanded scam-detection + transcription
facebook released social + ai marketplace features: users can create item collections, invite friends to group chats with sellers, and get ai-suggested questions to ask sellers
apple launched mini apps partner program, offering a reduced 15% commission rate for mini apps built with web technologies
uber is testing in-app driver video recording in india, to help deter rider and driver misconduct during rideshares
venmo released venmo stash, a rewards program offering up to 5% cash back for using the venmo debit card with select brands
threads released improved podcast features, including emphasized link displays and profile links for creators
google notebooklm released their own deep research feature
google released gemini on google tv, replacing google assistant and enabling natural voice navigation + recommendations
google photos released new ai editing + ai search features, including modifying objects in images, and adding new styles like renaissance portraits
other llm news
baidu released ernie 5.0, a foundation model for content generation (text, image, audio, video) that reportedly beats gpt-5 on document understanding
google deepmind released sima 2, a video-game-playing agent that can navigate 3d worlds
anthropic is releasing an open-source method for evaluating ai chatbot bias, scoring models on fairness for answers to political questions
idk tbh
waymo launched driverless robotaxi rides on freeways in sf, phoenix, and la. i love taking waymos in sf, but freeways feel pretty scary, ngl. people are just starting to warm up to autonomous cars, and i wish waymo would keep building confidence with nationwide non-freeway coverage before tackling higher-risk freeways. a fatal incident feels inevitable, and i’d hate for that to undo the trust they’ve been slowly building
visa is testing stablecoin payments, letting businesses send stablecoins to consumer wallets (targeting gig workers + creators). i’m incredibly bullish on b2b stablecoin payments, but i’m not sure i fully understand where visa fits in; a big benefit of stablecoins is not paying visa fees every time you transact
