DeepSeek's R1 stuns Silicon Valley

Read time: under 4 minutes

A popular joke in Silicon Valley goes something like this: America innovates, China imitates, and Europe regulates. As of today, the middle part of that joke is no longer true. China’s new DeepSeek R1 model has shocked Silicon Valley, and many long-held assumptions about Chinese innovation and AI have evaporated overnight. Some are calling it a hoax, while others are calling it a gift to humanity.

Extraordinary events demand extraordinary coverage. We’ll be departing from our usual coverage of the industry to focus mostly on R1 today.

Today’s Insights

• DeepSeek’s R1 challenges Silicon Valley’s AI dominance

• The Implications: The Good, Bad, and Ugly of R1

• Tutorial: How to automate job applications with ChatGPT Operator

• Everything else you should know today

• Memes and what’s trending on socials

TODAY IN AI

DeepSeek’s R1 challenges Silicon Valley’s AI dominance

Groundbreaking technologies usually come in two phases. First, invention — someone creates something new that didn’t exist before. Second, innovation — someone makes the invention cheap and reliable so that everyone can use it. 20 years ago there was no iPhone. Today, over 4 billion people use a smartphone.

AI models capable of reasoning already exist. But DeepSeek’s R1 appears to have delivered an innovation on a scale that puts a PhD-level AI reasoning model in billions of pockets around the world — something people in AI like to call “intelligence too cheap to meter.“

The best things in life are (almost) free. Not only does R1 match or exceed OpenAI’s o1 reasoning model across multiple benchmarks, it also delivers the output for approximately 5% of the cost. The cherry on top of the cake: the model is fully open-source. Yes, anyone in the world can use, copy, or fine-tune the model to their liking (assuming that they have the computing power to do it). And DeepSeek even hit #1 on Apple’s app store chart for free apps.

Denial, Despair, Delight. The release has caused a stir in Silicon Valley, to put things mildly. Some are calling it a Chinese psyop designed to undermine America, while others are calling it an astonishing technical achievement. Scroll down to the next section to get the full details on the controversies.

PRESENTED BY SPEECHMATICS

👂 Speechmatics - Introducing the Best Ears in AI

For many industries such as healthcare, education, or food service, Voice AI that understands most of the words they hear, isn't good enough.

For when precision is needed, Speechmatics offers:

✅ Ultra-accurate, real-time speech recognition, even in noisy environments

✅ Inclusive understanding of any language, accent, or dialect

✅ Seamless support for group conversations

Customer relationships are built on how well you listen. Speechmatics ensures your AI apps listen better than ever. Experience it in seconds.

FROM THE FRONTIER

The Implications: The Good, The Bad, and The Ugly

A look at the hot takes and bold claims in the aftermath of R1

Is R1 a Chinese psyop designed to defeat the US?

This view started to gain traction after entrepreneur and investor Avichal questioned DeepSeek’s claim that R1 only took $6M to train, arguing that it’s more likely that China has access to a deep bench of AI chips that they don’t want to reveal. Neal Khosla, founder of two venture-backed health tech companies, Curai and Koko, didn’t hold back any punches either. He accused DeepSeek of being a “CCP state psyop“ that is faking its low price in the hopes that “everyone switches to it to damage AI competitiveness in the US.”

Or maybe it genuinely is the real deal

No wonder venture capitalists are up in arms about a cheap open-source model. They’ve invested billions into closed model AI companies like OpenAI and Anthropic. Khosla’s post was community-noted with the clarification that there is no evidence to support his claims and that Khosla’s father is a major investor in OpenAI.

Whether or not R1 has some ulterior motive behind it, it’s hard to deny that it is a major technological breakthrough. Venture Capitalist Marc Andreessen called R1 “one of the most amazing and impressive breakthroughs I’ve ever seen — and as open source, a profound gift to the world.“ Former GitHub CEO Nat Friedman echoed a similar sentiment: “The deepseek team is obviously really good. China is full of talented engineers. Every other take is cope. Sorry.“

The implications:

• At first look, this is a threat to closed model companies like OpenAI and Anthropic, which have poured billions into AI training.

• But maybe not. If the best models are commoditized, then the companies with the most computing power and chips — like OpenAI — will get ahead.

• Fears about political censorship in R1 are valid but may be overblown. Since the model is open-source, users should be able to fine-tune it to their preferences.

• There’s also a geopolitical angle. The American chip export ban on China does not appear to be having its intended effect.

• Markets are also watching closely. Is this good news or bad for the investors that have poured billions into AI chips? And will AI companies be able to generate the revenues that justify these capital expenditures?

PRESENTED BY INNOVATING WITH AI

Want to become an AI Consultant?

Innovating with AI just welcomed 200 new students into The AI Consultancy Project, their new program that trains you to build a business as an AI consultant:

• Tools, frameworks, and a 6-month plan to build a 6-figure AI consulting business

Click here to request early access to The AI Consultancy Project

THE AI ACADEMY

How to automate job applications using ChatGPT operator

• Go to ChatGPT Operator and log in with your account.

• Now upload your resume, enter your prompt, and press enter.

Sample prompt: ‘Take my resume, attached, and search on [enter platform] to find me a job based on my experience. They should either be remote or in [enter location]’

• Operator will prompt you to log in to your account and search for jobs for you.

• You can interact with your browser within Operator to apply for jobs.

(NOTE: Google Operator is available to ChatGPT Pro subscribers only)

AI & TECH NEWS

Everything else you need to know today

📱 Siri’s Savior: Apple exec Kim Vorrath — a 36-year vet with a track record of launching iconic products like the iPhone and Vision Pro — is now being tapped to kickstart the company’s lagging AI initiatives, including a revamped version of Siri.

🎁 AI for All: Sam Altman confirmed that even free ChatGPT users will get a chance to try OpenAI’s state-of-the-art reasoning model, o3-mini, when it rolls out in the coming weeks. If you can’t wait that long, OpenAI just announced some handy updates to its new Canvas feature to hold you over.

✍️ Mental Gymnastics: The nonprofit Center for AI Safety created a new AI benchmark called Humanity’s Last Exam that’s so challenging no model has managed to score above a 10% on it.

📈 Markets: Twilio’s shares shot up 20.13% Friday after analysts predicted that AI startups would come to rely on the cloud company’s services for upcoming voice and messaging products. And traders are eagerly watching earnings reports from Meta, Tesla, Microsoft, and Apple scheduled for this week.

💰 Funding: ElevenLabs, a leader in AI voice cloning and dubbing, raised $250M in a Series C funding round. It’s now valued at around $3B. Retro Biosciences raised $1B after landing an exclusive partnership with OpenAI to extend the human lifespan. And LG acquired a majority stake in Bear Robotics, a Korean startup developing AI-controlled robot fleets.

PRODUCTIVITY

5 AI Tools to Supercharge Your Productivity

✅ Infuzu: Use a single platform that brings together all of the top-performing AI models and automatically finds the best one for every request.

✅ Telescope: Use an AI-powered research agent to search for leads with custom criteria, finding exactly who you need, no matter how niche.

✅ MoEngage Inform*: Save 95% Dev Bandwidth and 60% Costs as you manage Critical Alerts at Scale, with a Unified Platform for all your providers. Check it out →

✅ Tellers: Turn songs, voice recordings, podcasts, and more into captivating video edits in seconds.

✅ Roundups: Use AI to research products and generate guides that influence buying decisions.

^{* indicates a promoted tool, if any}

PROMPT OF THE DAY

Organize your notes

Prompt: [paste your notes here] here are some notes. I want to define my mission, vision, strategy, goals and roadmap. Can you help me reorganize them into this structure using bullet points?

Mission: What do we want to achieve?
Vision: What does the world look like after we've achieved it?
Strategy: What's our plan to achieve it?
Goals: How do we measure progress?
Roadmap: What do we need to build?

_{Source: Alex Banks / The Signal}

SOCIAL SIGNALS

What’s trending on socials today

💼 Career Hack: Midjourney founder David Holz shared a handy guide on how to land your next role in Silicon Valley.

🔍️ Search Star: App builder Sully Omar found that DeepSeek and websearch together outperform Perplexity for him.

⛰️ Steep Ask: There was a ‘Hike to stop AI’ on Sunday in SF. A16z VC Justine Moore is skeptical.

🏗️ In Progress: Sam Altman shared images of the Stargate Project in Texas.

📖 Demystifying Deepseek: This guide does a great job of explaining the R1 paper in detail for those who want to learn the model’s secrets.

Acquire new customers and drive revenue by partnering with us

Superhuman is the world’s biggest AI newsletter for businesses and professionals with 1M+ readers and 2M+ followers on socials working at the world’s leading startups and enterprises. Companies like Amazon, Hubspot, and Salesforce feature their products in Superhuman. You can learn more about partnering with us here.

Your opinion matters!

Got more feedback or just want to get in touch? Reply to this email and we’ll get back to you.

Thanks for reading.

Until next time!

Zain & the Superhuman AI team