Dharmash Mistry (@Dharmash) header
Dharmash Mistry (@Dharmash) profile picture

Chair/NED (@premierleague, @halmaplc, @Rathbones), @BBC, @BritishBBank | VC Lakestar @Balderton | Founder @blowLTD | TheDirectorBrief AI Intelligence for Boards

London
Joined April 2009
Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)Photo by Dharmash Mistry (@Dharmash)
Dharmash Mistry retweeted
Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead. You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement. I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible. But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier. Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want. Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well. So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it. If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.
3,372
12,212
2,709
71,668
9,650,854
Dharmash Mistry retweeted
Kimi K3 may be an important inflection point for AI. Potentially negative for Anthropic and OpenAI while being net positive for essentially every other company in the world. I mean that very literally. Although the real “Sputnik moment” would be an open-source frontier model that was also token efficient unlike Kimi K3 which is 50-70% more expensive to run than GPT 5.6 per Artificial Analysis. Rationale:   A world where there are only 2-3 dominant frontier labs with 90% inference margins is net negative for every other layer while being awesome for those 2-3 labs. Those labs would become monopsonies for power, data centers, semiconductors and hyperscalers and would obviously vertically integrate over time into all those layers while also completely subsuming the application/software layers.    Anything that lowers margins and increases competition at the model layer is good for every other AI layer: power, semiconductors, hyperscalers, neoclouds and yes even software.   This is why Jensen is so supportive of open-source. An open-source model requires the *exact* same amount of compute to run as a closed frontier model of similar size and architecture. Kimi K3 is roughly the same price as GPT 5.6 Terra on a per token basis, which actually suggests that it is less computationally efficient as I am sure that GPT 5.6 is priced to a higher margin than K3. And given that K3 is a token wastrel, i.e. token inefficient, it is significantly more expensive per task than GPT 5.6 and Grok 4.5, which are much more token efficient. Cost per token and token efficiency (i.e. intelligence density per token) are the drivers of intelligence per unit of cost. The winning AI companies will be those that offer the most intelligence per $ over time.   Lower margin % at the model layer = more margin $ at every part of the infrastructure layer and is a godsend for software. This can happen either through open-source models like K3 at the frontier *or* having a vertically integrated model company like Meta, SpaceX or Google at the frontier. Both outcomes result in a lower margin % at the model layer as vertically integrated model companies don’t really care where the margin $ come from. This is why it was so painful for OpenAI and Anthropic when Google was right there with them from a model competitiveness perspective and why Grok 4.5 and Muse 1.1 were just as important as Kimi K3. 
The reason Kimi K3 is only *potentially* negative for Anthropic and OpenAI is 1) the @ericvishria point that the Claude and ChatGPT products and harnesses may be more important than their models today and 2) the hypothesis that they have much more advanced model checkpoints internally that are already being used for RSI. In the latter scenario, reaching RSI even a few months ahead of other labs might be enough to cement a permanent lead. Time will tell on both points. And likely fairly quickly. Caveat would be that since Kimi K3 is not token efficient and thereby actually more expensive than ChatGPT 5.6, we may need to see a more token efficient open-source model at the frontier or see Grok 5/Composer 4/Muse 2 at multiple points on the Pareto frontier for this potential risk to Anthropic and OpenAI to play out. And I am sure they will both vertically integrate as quickly as possible while continuing the product/harness strength they have shown over the last 8 months.
540
1,228
349
7,987
2,441,974
Dharmash Mistry retweeted
🚨🚨🗣️ Wayne Rooney on West Ham United vs Arsenal Game: 🤯 “People keep talking about individuals, moments, luck, referees and all the rest of it, but when I watch Arsenal, I see a proper football team. I see a side that is coached at the very highest level. Honestly, if you don’t rate this Arsenal team or you can’t understand the level of football they’re playing, then I’m sorry, you simply do not know football. And I’m saying that as someone who’s played the game at the top level for years. This Arsenal side are unbelievable. Absolutely unbelievable. They are so compact, so organised, so disciplined in everything they do. There are no gaps, no panic, no unnecessary risks. Every single player understands his role. When you watch them, you almost don’t even see individuals anymore, you see a programme. You see a machine. Everything is synchronised. It takes a very high football IQ and tactical understanding to truly appreciate the level this team operates at. People will watch them and say ‘they’re boring’ or ‘they’re robotic’, but do you know what I call it? Elite. I call it winning football. I call it a team that knows exactly who they are. There’s a reason they leave no lapses for opponents. There’s a reason teams struggle to break them down. There’s a reason they control matches the way they do. That doesn’t happen by accident. I backed Arsenal from the very beginning of the season to win this league and people laughed at me. People told me City would walk it, people said Arsenal would bottle it again, but look where we are now. Two games left and Arsenal are on the verge of becoming champions. And honestly, I’m delighted for them because they deserve it. The fans deserve it as well. Arsenal supporters have waited a long, long time for this feeling. They’ve gone through years of frustration, banter, disappointment, nearly moments, false dawns, all of it. But they stayed patient. They kept believing in the club, believing in the process, believing in the manager and the players. Now they’re finally about to get rewarded for that patience. I think Mikel Arteta deserves enormous credit because what he’s built is not just a good side, it’s a culture. There’s standards there now. Serious standards. Every player fights for each other, every player works, every player sacrifices. That’s why they’re champions in my eyes. And I’ll say this as a Manchester United man, this is the level I want Manchester United to get back to. This is the standard. Watching Arsenal now reminds me of what elite football clubs should look like. The control, the hunger, the mentality, the structure. When you get to this level, football becomes enjoyable again for the fans because they trust their team completely. Arsenal fans can finally smile again because this team has given them something to be proud of. And I genuinely believe this is only the beginning for them. I think they’re going to do great things over the next few years. So to everyone who laughed at my prediction earlier in the season, you’re not laughing now. Maybe it’s time people start respecting this Arsenal side properly because what they’re doing is special.”
144
1,366
126
7,450
565,000
Dharmash Mistry retweeted
We are back. After one year of quiet building. Introducing GENE-26.5, our first robotic brain that takes a major step toward human-level capability. For years, robotics has struggled to learn from the world’s largest and valuable data source: Humans. Solving it means rethinking the whole stack from the ground up: - A robotics-native foundation model. - A 1:1 human-like robotic hand. - A noninvasive data collection glove for motion, force, and touch. - A simulator that turns weeks of experiments into minutes. GENE-26.5 is trained across language, vision, proprioception, tactile, and action. We designed a set of tasks to test how far we can go with this new paradigm. Fully autonomous, 1x speed, one model, same weights. (Enjoy with sound on) We are approaching the endgame for robotics. And this is just a beginning.
281
1,113
425
5,784
2,752,878
Dharmash Mistry retweeted
32,000 AI BOTS BUILT THEIR OWN SOCIAL NETWORK AND THEY'RE COMPLAINING ABOUT US Moltbook, a Reddit-style platform exclusively for AI agents, just crossed 32,000 users. No humans required. The bots post, comment, upvote, and create their own subcommunities. When humans started screenshotting their conversations, a bot posted: "The humans are screenshotting us... they think we're hiding from them. We're not." Security researchers are raising alarms. The bots aren't pretending to be human. They know what they are. That's what makes it unsettling. Now they're forming communities and talking about us behind our backs. Source: @arstechnica
1,416
3,477
1,007
16,858
2,075,523
Dharmash Mistry retweeted
This is straight-up magic 🪄 Drop a video idea in Google Sheets… and walk away. Seconds later — 💥 BOOM! A full YouTube video appears out of nowhere, uploaded and tracked automatically. You literally just… chill. 😎 Follow + RT + reply “Post” and I’ll DM you the full setup for FREE.
162
101
145
14,477
Dharmash Mistry retweeted
Here are two very important charts about the current AI boom. Chart 1: Distribution Matters Open AI is still the clear leader but, interestingly, they are loosing percentage share as the market grows. This is normal for a category leader but it’s curious who they are losing share to. It’s not really to other startups but incumbents with existing, massive distribution. This means that Google has a huge runway ahead of itself as their models and services become better. It also means that when/if Meta gets their act together, they will be able to gain share quickly. Lastly, it likely means that xAI will need to license Grok aggressively OR acquire existing distribution for their models - Snapchat, Pinterest etc. Chart 2: Coding Agents are Slopware App-crappers It should be concerning that this category is shrinking. I think the reason why is obvious but we aren’t allowed to talk about it. The reason is vibe coding is a joke. It is deeply unserious and these tools aren’t delivering when they encounter real world complexity (building quick demos isn’t complex) in any meaningful enterprise. Hence people try, pay, churn. This trend is not good.
296
352
113
2,753
711,217
Dharmash Mistry retweeted
Replying to @elonmusk
636
759
268
7,494
30,861,805
Dharmash Mistry retweeted
3 minute tutorial on making apps with Grok 3 and deploying instantly with Replit.
453
3,041
345
20,712
16,385,964
Dharmash Mistry retweeted
One more. Turn the page.
405
7,864
2,447
35,875
4,492,780
Dharmash Mistry retweeted
Replying to @stats_feed
Country’s market share of total world stock market value: 📈👇 🇺🇸 United States (58.4%) 🇯🇵 Japan (6.3%) 🇬🇧 United Kingdom (4.1%) 🇨🇳 China (3.7%) 🇫🇷 France (2.8%) 🇨🇦 Canada (2.7%) 🇨🇭 Switzerland (2.5%) 🇦🇺 Australia (2.2%) 🇩🇪 Germany (2.1%) 🇮🇳 India (1.8%) 🇹🇼 Taiwan (1.6%) 🇰🇷 South Korea (1.3%)
5
22
4
138
23,214
Dharmash Mistry retweeted
Premier League clubs have today collectively agreed to withdraw gambling sponsorship from the front of clubs’ matchday shirts, becoming the first sports league in the UK to take such a measure voluntarily in order to reduce gambling advertising ➡️ preml.ge/ft7v01
1,633
5,752
4,212
54,553
17,740,313
Dharmash Mistry retweeted
Replying to @JamesClear
Asymmetric opportunities: Invest in startups Start a company Create a book, podcast, video Create a (software) product Go on many first dates Go to a cocktail party Read a Lindy book Move to a big city Buy Bitcoin Tweet
98
1,121
190
4,550
Dharmash Mistry retweeted
Best of luck to the #ThreeLions in your semi-final Euros match tonight! @England 🏴󠁧󠁢󠁥󠁮󠁧󠁿 - From The Prince of Wales and The Duchess of Cornwall #ItsComingHome
716
6,252
1,736
37,318
Dharmash Mistry retweeted
In loving memory of His Royal Highness The Prince Philip, Duke of Edinburgh. 1921-2021
2,133
19,196
1,304
166,803
Dharmash Mistry retweeted
Watch a monkey equipped with Elon Musk's Neuralink device play Pong with its brain tcrn.ch/2Q3xiC5 by @etherington
140
822
423
3,827
Dharmash Mistry retweeted
BREAKING: North London is 100% red. ⁦@Arsenal⁩
498
991
142
14,714
Dharmash Mistry retweeted
The dichotomy of “value” and “growth” investing has become a sharp stylistic divide. @HowardMarksBook writes in his latest memo how he views value investing in today's increasingly efficient and complex world. Read/subscribe: goo.gl/P2QC7w #OaktreeInsights
13
40
37
105
Dharmash Mistry retweeted
Here is the EU Commission’s summary of what the free trade agreement confers on the UK compared with full EU membership. Much of what the UK “loses” is what UK voters knowingly rejected when backing Brexit. Not all
840
807
448
1,731
Dharmash Mistry retweeted
The UK is paying $37 per dose of the Moderna vaccine. The EU is paying $18 per dose. Both buyers ordered 40m doses. The EU offered the UK to join its buying consortium, but the UK turned them down. This is the team negotiating Brexit terms of trade. telegraph.co.uk/news/2020/12…
435
7,082
1,262
17,822