Alexandru Voica
banner
alexvoica.com
Alexandru Voica
@alexvoica.com
Corp affairs at Synthesia, advisor at MBZUAI

Guns N' Roses, basketball, Philip K. Dick, intelligent machines, peanut butter + maple syrup are a few of my favorite things.

https://linktr.ee/alexvoica
Michael Burry is betting $1bn against Palantir because he sees them as a proxy for the AI bubble.

What he fails to understand is that the next leg of the AI cycle will be decided in the boiler rooms of the enterprise: www.computerspeak.co/p/palantir-i...
Palantir is the world's must successful forward-deployed engineering company
The future of ads is AI personalization; Apple Watches use AI to detect heart damage; free from OpenAI, Microsoft charts their own AI destiny; AI data centers are warping the US economy
www.computerspeak.co
November 7, 2025 at 2:01 PM
My conversation with Bernard Marr and Sasha Rubel from AWS about AI adoption and innovation is now live on YouTube: youtu.be/HbaEH2bU800
Mind the Gap - Navigating The AI Innovation Frontier
YouTube video by Bernard Marr
youtu.be
November 5, 2025 at 12:55 PM
In an article for Transformer, I argue (based on lessons learned at @synthesia.io) that drafting lengthy applied AI strategies and coming up with new pilots/sandboxes is not the way forward if we're serious about AI adoption:

transformernews.ai/p/audits-not...
Audits, not essays: How to win trust for enterprise AI
Opinion: Alexandru Voica argues that application-layer AI companies are best off opening themselves up to rigorous testing rather than opining on AI safety
transformernews.ai
October 29, 2025 at 8:12 PM
Are you a company looking to adopt AI but unsure how to navigate Europe's complex AI regulations? 🇪🇺

Join me for @synthesia.io's webinar on November 5th as I talk to experts who are making great-tasting lemonade out of the lemon that has been the EU AI Act: synthesia-io.zoom.us/webinar/regi...
October 28, 2025 at 10:29 AM
I had an amazing chat with @bernardmarr.bsky.social and Sasha Rubel from @awscloud.bsky.social about how organizations can speed up AI adoption, how we can scale infrastructure and build regulatory regimes that favor responsible innovation, and deploy AI agents across multiple industries.

#AWS
October 23, 2025 at 8:14 AM
New data from @nathanbenaich.bsky.social's State of AI Report released this week shows how AI adoption is doing something unfashionable for a bubble: it’s compounding.

www.computerspeak.co/p/the-2025-s...
The 2025 State of AI Report shows an industry maturing; business leaders learn to live in the AI bubble; Sora 2 reignites AI slop debate; how AI became personal assistants; updates from OpenAI DevDay
A startup aims to create a new DeepSeek moment; ASML exec slams EU AI overregulation; AI is now in every industry; AI toys and sexy chatbots come to America; how AI is changing white-collar work
www.computerspeak.co
October 10, 2025 at 4:16 PM
This week, it’s become painfully obvious that OpenAI, YouTube, TikTok and Meta are all gunning for the same prize: the default feed for AI native video.

Here's why: www.computerspeak.co/p/ai-video-c...
October 3, 2025 at 3:21 PM
Synthesia 3.0 is our new AI video platform reimagined for the future of video.

Victor Riparbelli went on Bloomberg Technology to chat with Ed Ludlow and Caroline Hyde about @synthesia.io's new products and the benefits they will bring to businesses: youtu.be/t319kg2yT5A
AI Video Clone Startup Launches New Tools
YouTube video by Bloomberg Technology
youtu.be
October 2, 2025 at 9:13 AM
It's very clear that Meta's Vibes launch was rushed to pre-empt the Sora 2 release.

The strategy backfired because it attracted a wave of people predictably dumping on Vibes and AI video as a social experience, clearing the way for Sora to get a more neutral reception.
September 30, 2025 at 9:18 PM
More than 10,000 people have signed up for our LinkedIn event on October 1 to learn how @synthesiaIO will change video as a format, and what's coming next for conversational agents that are built to work in the real world and deliver business value:

lnkd.in/esjikxvp
September 29, 2025 at 4:11 PM
Congratulations to Germany's Minister of State for
@bundeskultur.bsky.social Wolfram Weimer for launching Weimatar – the first AI avatar of a German government member created with @synthesia.io

More information here about the project: kulturstaatsminister.de/erster-virtu...
September 24, 2025 at 12:28 PM
I answered a few questions for the Friday Q&A section in @politico.eu's Morning Tech UK newsletter.
September 23, 2025 at 6:53 AM
AI grifters such as Eliezer Yudkowsky or Ed Newton-Rex, once dressed in TED-stage gravitas and penning op-eds for The Guardian, are now pivoting to the more niche corners of the MAGA mediasphere:

www.computerspeak.co/p/the-ai-gri...
The AI grift, a tragedy of mainstream media failures in three acts; the US and UK sign tech deal; Google's plan to generate more AI training data; meet the new AI startup whisperer
Why AI won't replace junior engineers; the AI movie factory is ramping up; Hangzhou is China's answer to Silicon Valley; life at Lovable; how people use OpenAI and Anthropic's chatbots
www.computerspeak.co
September 19, 2025 at 11:59 AM
For me, generative AI’s best trick isn’t replacing people, it’s compressing the cost of quality.

In other words, there’s a meaningful line between influencers creating with AI and influencers who are AI:

www.computerspeak.co/p/you-cant-y...
You can't (yet) fake influence; Anthropic's approach to red teaming; Synthesia's AI clones are more expressive than ever; Chinese companies still want Nvidia's chips; meet the AI gambling agents
AI is outperforming human recruiters; AI is getting cheaper and more expensive at the same time; AI labs struggle to keep chatbots from talking about suicide; how AI is affecting software engineering
www.computerspeak.co
September 5, 2025 at 1:55 PM
In an in-depth feature for MIT Technology Review,
@rhiannonwilliams.bsky.social writes about (and becomes one of!) @synthesia.io's new Express-2 avatars:

technologyreview.com/2025/09/04/1...

📰 @technologyreview.com | tip @techmeme.com
Synthesia’s AI clones are more expressive than ever. Soon they’ll be able to talk back.
The uncanny valley is narrowing. Are we ready for what comes next?
technologyreview.com
September 4, 2025 at 11:17 AM
Spoke with @mshannahmurphy.bsky.social at the @financialtimes.com about the rise of AI influencers: ft.com/content/cb85...

I believe gen AI is going to be a great equalizer across many industries, particularly in the creative industry where 𝒘𝒉𝒐 you know can be just as important as 𝒘𝒉𝒂𝒕 you know.
The rise of the AI influencer
Computer-generated profiles are gaining traction with brands but may pose a threat to real life endorsers
ft.com
September 1, 2025 at 7:19 PM
NEW: I spoke to three of the winners of the International Olympiad in AI, an annual competition for high schoolers that took part earlier this month:

www.computerspeak.co/p/inside-202...
Talent won over teraflops at the 2025 International Olympiad in AI; data center investments are showing up in GDP data; the AI bubble is fuelled by ARR but doomed by churn; meet the AI hype house
Young grads are making hundreds of thousands with AI; Agents4Science is a scientific conference powered by AI; how AI startups can hire or retain staff in the talent war; inside DHL's AI upgrade
www.computerspeak.co
August 29, 2025 at 1:32 PM
Just out on @bloomberg.com: @synthesia.io CEO Victor Riparbelli sits down with Tom Mackenzie to discuss how AI is reshaping the workplace, including through products such as AI dubbing or AI avatars designed to make video more engaging, personal, and effective at scale: youtu.be/nxgpCr9oULY
August 18, 2025 at 1:00 PM
This week, the UK government suggested that deleting old emails can help address water shortages.

Instead of wasting hours on performative actions, spend five minutes reading about how AI is affecting water consumption:

www.computerspeak.co/p/is-ai-real... 🌊💻
Is AI really draining our water supply? Alan Turing Institute on the brink of collapse; Meta's struggles to build its AI team; how top chatbots change your mind; Amazon ventures into neurosymbolic AI
Meta AI allowed "sensual chats" between its chatbots and children; Mistral sets its sights on the Middle East; companies are still struggling to see the ROI of AI; China's lead in open source AI
www.computerspeak.co
August 15, 2025 at 4:47 PM
Sam Altman described "a model that continuously learns as it's deployed from the new things it finds" something that to him "feels like AGI."

In this week's Computerspeak, I attempt a more detailed spec for what AGI (or superintelligence) could mean: www.computerspeak.co/p/a-definiti...
A definition of superintelligence for the real world; Nvidia warns against GPU "kill switches"; Harvard and MIT students are leaving university because of AGI; Demis Hassabis on the future of AI
Business Insider profiles Cloudflare's CEO; Google introduces new Genie 3 world model; Why Anthropic has an edge in the AI talent war; Trump admin's pitch to Asian countries: don't over-regulate AI
www.computerspeak.co
August 8, 2025 at 12:13 PM
Introducing AI Dubbing in @synthesia.io 🎉

Now you can dub your existing videos into 30+ languages while keep the original speaker’s voice, with perfect lip-sync — all done in minutes.

And for full and safe control over machine translations, we've built a Secure Editing feature.
July 25, 2025 at 12:49 PM
In this week's newsletter, I'm an old man yelling at the clouds about AI products designed by the socially disengaged, policed by the epistemically fragile, and deployed to users who mostly just want a powerful assistant and not an anime AI girlfriend:

www.computerspeak.co/p/are-we-bui...
Are we building machines to replace human connection? ChatGPT reaches 900 million downloads; chatbots in the classroom; Nvidia CEO travels to China; AI is reshaping Salesforce;
Grandmaster beats ChatGPT at chess without losing one piece; the AI startups focused on drug discovery; OpenEvidence has built the ChatGPT for doctors; inside Helsing's race to build AI for defense
www.computerspeak.co
July 18, 2025 at 10:05 AM
I spoke with two researchers from @mbzuai.bsky.social about their #ICML2025 paper showing that with $10 of cloud computing and an open weights model, watermarking can be easily tampered with:

mbzuai.ac.ae/news/challen...
Challenging the promise of invisible ink in the era of large models
MBZUAI's Nils Lukas and Toluwani Samuel Aremu presented research at ICML that highlights the need for more robust digital watermarking.
mbzuai.ac.ae
July 17, 2025 at 2:37 PM
In this week's newsletter, I compare Grok 4 to a vehicle with Level 2 autonomy (it's a good reasoning model), which of course is incredibly impressive compared to cruise control (GPT-2), but should not be confused with a Level 5 self-driving car (AGI):

www.computerspeak.co/p/elon-musk-...
Elon Musk tries to sell us Grok 4 in a lab coat; tech platforms are struggling with AI video; Amazon deepens Anthropic investment; Meta is trying to win the AGI race with money
Games are the new battleground for actors and AI; the most popular AI coding tools for engineers; inside the AI scraping fight that is changing the web; India scrambles to achieve AI independence
www.computerspeak.co
July 11, 2025 at 9:09 AM
Today, any child in Europe could have the opportunity to access better education in the form of a conversational AI model that is truly helpful.

Except the AI Act limits the use of this technology in education.

We should #StopTheClock on the AI Act: computerspeak.substack.com/p/on-ai-euro...
On AI, Europe must choose: become a factory or a museum; Runway ventures into world models; SAP CEO calls for more applied AI; ElevenLabs lays out IPO plans; China is winning in open source AI
Trumpworld is divided over AI regulation; OpenAI sees a brain drain to Meta; a startup built a hospital in India to test its software; digital workers arrive in banking; Hollywood's pivot to AI video
computerspeak.substack.com
July 4, 2025 at 9:08 AM