Ziad Matar (@wirelessly) header
Ziad Matar (@wirelessly) profile picture

πŸ•΄πŸ» parallel entrepreneur πŸƒπŸ½ accidental athlete πŸ” chief BBQ officer 🍸 amateur mixologist πŸ€“ geek β€’β€’β€’ all views expressed are mine

Global
Joined March 2009
Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)Photo by Ziad Matar (@wirelessly)
Someone asked me on Wednesday what was the biggest misconception about AI, I gave 2: that AI is free, and that good AI is very expensive. Both hide the same elephant in the room: nobody can know what AI will cost before running it.
1
1
2
91
The budget moves with the model chosen, the size of the input, the kind of output, how long the model reasons, caching, batching, volume steps and the mix of cheap and frontier models picked task by task.... and then there are agents doing what agents do!
1
15
That leaves a blank variable in 3 equations essential for any business: profitability, return on investment and pricing. Read all about the missing variable in every AI business case, and why we built @TokenandoAI , in this week’s issue of The Compute: tokenandoai.substack.com/p/t…
1
28
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Friday November 6, 2026 / Wall Street suffers its sharpest single-session loss since March 2020, with the Nasdaq 100 falling 6.4% and erasing $2.3tn as Democratic midterm gains trigger a sweeping AI infrastructure sell-off / Two fund managers argue the AI equity trade broke because investors bundled infrastructure spend, model capability, and enterprise revenue into a single return thesis that the market is now pricing separately / Asian semiconductor markets absorb the Wall Street shock, with South Korea's Kospi triggering a circuit breaker and Bloomberg Economics warning that tighter US export controls could expose China's AI-driven export growth / Silver Lake and Saudi Arabia's Public Investment Fund approach Unity Software at a reported $25bn valuation, following their $55bn take-private of Electronic Arts in August 2026 / SemiAnalysis publishes GDPx, a real-world agent benchmark across 206 professional tasks, with the leading model scoring 12% and every model scoring zero on finance and legal work under binary grading / OpenAI releases GPT-6 Cyber to approved security customers, scoring 48.5% on ExploitGym and developing 948 working Linux kernel exploits at roughly $1,279 each, against historical bug-bounty rates of $20,000 to $70,000 / US voters pass 39 of 40 local ballot measures restricting data centre construction, with an average 75% vote share, as Democratic candidates win governorships in Texas, Georgia, and Ohio on moratorium platforms πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
2
1
2
110
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Thursday November 5, 2026 / China's NDRC blocks SoftBank's $5.4bn acquisition of ABB's robotics division, citing risk that operational data could train OpenAI's AI models / Toyota establishes Toyota Robotics as a company-wide unit from January 2027, targeting 400,000 factory robots and up to $6.4bn in annual spending from 2028 / OpenAI acquires Pierre Computer Company to fold its code-storage and developer tools into the Codex platform, targeting GitHub's infrastructure position / Apple cuts daily AI spending caps for engineers from $300 to $200, with further reductions to $100 and $25 under consideration across non-engineering teams πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
2
1
1
38
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Wednesday November 4, 2026 / Texas elects its first Democratic governor since 1990 on a platform opposing new data centre construction, putting more than 25% of planned US capacity at risk / Nvidia allocates just 1% of a $70bn Chinese RTX chip order, with global 2027 production of 3 million units already exceeded by Chinese demand alone / OpenAI reports GPT-6 Astra designed a physical hardware device in under a week, though manufacturing validation required multiple rounds of human review / Canva cuts 1,271 jobs and sees its valuation fall from $42bn to $31bn in 6 months, with AI rollout problems linked to the restructuring / Anthropic releases FlashProve, an open-source cryptographic system for verifying which AI model produced a given output, tested at 110 tokens per second on a single H100 / Data centre operators deploy converted jet engines as primary power, with grid connections in Northern Virginia now taking approximately 7 years versus a 24-month turbine deployment timeline πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
1
1
1
76
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Tuesday November 3, 2026 / Anthropic files amended IPO prospectus revealing $16.96bn in Q3 revenue, a $895bn infrastructure commitment, and a $23bn headline loss driven largely by accounting adjustments / Morgan Stanley chief executive Ted Pick puts cumulative AI capital expenditure at $10tn to $15tn over the next 12 years, with $2tn to $3tn already deployed / Goldman Sachs restricts Claude Code and GitHub Copilot to engineering staff, withdrawing the tools from traders and bankers after months of broader access / Abu Dhabi Crown Prince witnesses signing of 4 OpenAI agreements spanning ADNOC, Mubadala, the Abu Dhabi Government, and Inception42 at the ADIPEC conference / Apple warns of further gross margin compression in the December quarter as memory chip costs reach levels chief executive John Ternus describes as unprecedented / OpenAI appoints former US Army Secretary Dan Driscoll to a newly created Senior Counsellor for National Security Modernisation role πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
1
1
1
61
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Friday October 30, 2026 / Alphabet, Amazon, Microsoft, and Meta report Q3 earnings as $750bn in AI capex collides with slowing advertising and e-commerce growth / DeepSeek closes a $14bn to $16bn funding round at a $75bn valuation, with Chinese state-owned investors taking a 27% share / Intel shares fall more than 20% after Q4 guidance misses analyst estimates by roughly 9%, despite 78% data-centre revenue growth / OpenAI publishes an updated Model Spec reorganising intended model behaviour around 5 core values, acknowledging current models do not yet fully reflect it / DOJ examines Nvidia's $12.93bn planned acquisition of Hugging Face, asking whether it would give Nvidia control of a fundamental AI model distribution point πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
54
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Wednesday October 28, 2026 / Bloomberg Intelligence projects China will lead in large language models and 7 other technology categories by 2030, with top Chinese models now within 3% of US benchmark scores / OpenAI in discussions to buy Google TPUs as Oracle becomes the first outside cloud provider to offer the chips, with Anthropic's IPO prospectus disclosing up to $275bn in combined TPU commitments / SK Hynix posts record quarterly revenue of 104.6 trillion won ($78.9bn) and an 80% operating margin, driven by HBM4 shipments and a 632% year-on-year profit surge / Anthropic and the US General Services Administration announce a OneGov agreement giving government agencies a 40% discount on Claude token usage, replacing a deal set to expire October 31 πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
39
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Tuesday October 27, 2026 / Anthropic files its S-1 with the SEC, reporting $16.3bn in first-half 2026 revenue and a $780bn to $810bn compute commitment, targeting a Nasdaq listing under the ticker ANT / OpenAI releases GPT-6.1 Terra at $1 per million input tokens, half the price of its predecessor, with hallucination rates cut from 88% to 38% / Scale AI opens its Dubai hub at DIFC, its 3rd Gulf market in under 2 years, as Dubai AI Campus raises its 2030 target to 1,000 companies and 5,000 jobs / Anthropic's 7 co-founders include an ad-free pledge for Claude in the IPO prospectus, formalising a position first stated in February 2026 / SpaceX commits to a $40bn GPU finance lease with Apollo Global Management, with BlackRock, KKR, and Goldman Sachs Asset Management as anchor investors / OpenAI claims 3 AI-generated scientific results in mathematics, computer science, and physics, introducing an external academic shepherd framework for verification πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
107
the concentration of the top 10% of firms accounting for 99.5% of inference spend, shows that the AI deployment wave is still a narrow phenomenon concentrated in venture-backed companies and a handful of large enterprises...
Replying to @TokenandoAI
Torsten Slok, chief economist at Apollo Global Management, published a note on October 24, 2026, arguing that AI has delivered zero measurable financial return to US companies outside the technology sector. Slok analysed what he calls the S&P 493, the S&P 500 minus Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla, and found that revenue per worker and profit margins are flat, with no evidence of AI-driven productivity improvement. By contrast, revenue per employee at the 7 largest technology companies has risen from approximately $550,000 in 2022 to approximately $780,000 in 2026, and their profit margins have climbed to nearly 30%, against approximately 10% for the S&P 493. Slok also found that AI spending is highly concentrated: the top 10% of firms account for 99.5% of model serving and inference spend, and 80.4% of the top 1% of AI spenders are venture-capital-backed. The Ramp AI Index shows 55.9% of US businesses paying for AI in September 2026, but the median spend is $15.70 per employee against $9,567.53 for the top 1%. Slok concluded that mainstream corporate America is beginning to treat token spending as a cost centre rather than a growth driver. Apollo is a major lender and investor across the technology sector; its chief economist's view on AI returns will influence credit decisions and portfolio positioning across AI-dependent businesses that have not yet demonstrated financial returns. πŸ€– made with AI Full details via Apollo Global Management / apollo.com/wealth/insights-n…
15
good move from Nvidia to open-sourcing of Flora which reduces the cost of entry for the robotics ecosystem while keeping developers on Nvidia hardware...
Replying to @TokenandoAI
Nvidia's GEAR research lab has open-sourced Flora, a suite of 3 fine-tuned Qwen3.5 models designed to generate the labels required to train robot foundation models: timestamped action captions, process rewards, and outcome rewards. The system is already running in Nvidia's internal pipeline, labelling up to 200,000 hours of video per day across 512 GB300 GPUs, and is used to train Nvidia's GR00T robot models. In a 50-video test, Flora achieved 57.1% annotation quality at an estimated cost of $0.72 per video hour, compared with 53.7% quality at $88.49 per video hour for GPT-6 Astra at maximum reasoning effort. The $0.72 figure is a projection based on an assumed GPU rental price of $10 per hour, not a measured production invoice; additional costs including storage, engineering time, and review labour are excluded from the calculation. Flora's strongest demonstrated advantage is failure detection within episodes, where it recalls 53.3% of failed steps against 0% for GPT-6 Astra. GPT-6 Astra leads on robot video captions and rubric-based scoring. Flora is available on Hugging Face under an Nvidia non-commercial licence, which prohibits commercial redistribution. The Apache 2.0 licence on the underlying Qwen3.5 base models does not extend to the fine-tuned Flora weights. Nvidia positions Flora as an open alternative to commercial services such as Scale AI's Physical AI Data Engine. πŸ€– made with AI Full details via The Decoder / the-decoder.com/nvidia-open-…
22
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Monday October 26, 2026 / Anthropic files its S-1 with an $80.7bn first-half 2026 net loss and $518bn in committed cloud obligations, as co-founder Jack Clark moves his recursive self-improvement forecast forward to 2027 / Late-delivery penalties in AI data centre contracts reach $5m per day, triggering a specialist insurance market and raising the probability that penalty clauses will be triggered on most current projects / Omani investment group Mubbadrah leads a $3bn, 400MW AI data centre consortium announced during Sultan Haitham bin Tarik's state visit to the United Kingdom, with Crusoe named as a potential off-taker / Brookfield Asset Management opens a Doha office anchored by its $20bn joint venture with Qatar's AI subsidiary Qai, adding Qatar to its sovereign wealth fund relationships across Kuwait and Saudi Arabia / Nvidia open-sources Flora, a robot video labelling suite that achieves 57.1% annotation quality at $0.72 per video hour against $88.49 for GPT-6 Astra, under a non-commercial licence / Apollo chief economist Torsten Slok finds flat revenue per worker and unchanged profit margins across the S&P 493, concluding that AI has delivered zero measurable financial return outside the technology sector / Anthropic's Applied AI team publishes guidance recommending that enterprises consolidate multi-agent architectures into a single agent loop, citing a 2x latency improvement in documented enterprise comparisons πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
2
111
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Friday October 23, 2026 / OpenAI closes in on a $30bn funding round at a $1.25 trillion valuation, with UAE investors and Coatue leading a deal that pairs equity with a compute supply agreement / AI-linked equities shed more than $1 trillion in value across the week, with CoreWeave falling 10.1% on October 23 as markets reprice debt-financed infrastructure risk / HUMAIN expands its Nvidia collaboration to 84 megawatts of AI compute, with 50MW of Vera Rubin capacity reserved for an unnamed US client / Gartner survey of 2,002 CIOs finds only 13% report significant or transformational value from AI, while the average number of use cases rose 400% in a single year / Salesforce plans to charge enterprise customers for external AI agents including Claude operating within its applications, with registration requirements arriving as early as November 2026 / Google announces Agent Compute Engine at AGNTCon, giving each AI agent its own hardware-isolated microVM with a zero-secrets security architecture / Al-Futtaim survey of 1,143 GCC professionals finds 82% use AI weekly yet only 20% have received extensive employer training, with frontline workers the most underserved group πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
68
€94,749.99 blinked on my neighbourhood restaurant’s card terminal last Friday. Not a number the place sees in a quarter. The beer distributor had left his machine at the bar and taken the restaurant’s by mistake. A day of lunches went to the wrong account. Nobody meant any harm.
1
2
50
Same week: AI found a decade-old flaw in the software behind 70% of the world’s websites before the bad guys did. The fix was ready to be launched today. Then someone published a demo video by mistake and handed attackers a head start.
1
20
Whenever someone asks me whether AI will make it easier for hackers to move money out of our bank accounts, my answer is always the same: the problem is usually the human in the loop. The Compute by @TokenandoAI Issue XXV is out: tokenandoai.substack.com/p/t…
πŸ€– Made with AI
1
1
32
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Thursday October 22, 2026 / Demis Hassabis compresses his AGI timeline to 2 to 3 years, converging with Anthropic's Dario Amodei on the same window / Google Cloud adds OpenAI's GPT-6 models to its platform, becoming the only hyperscaler offering Gemini, Claude, and GPT under one invoice / Blackstone's Jas Khaira argues AI infrastructure debt fears are overstated, even as CoreWeave bonds trade below 80 cents on the dollar / AT&T and OpenAI formalise a legal AI partnership, with AT&T signalling its external law firm panel could shrink from 30 firms to around 18 to 20 / Ericsson ConsumerLab finds 63% of UAE smartphone users expect to delegate daily tasks to AI agents by 2030, double the global projection / BlackRock's Larry Fink warns that a slow data-centre buildout carries its own inflation risk, in direct tension with central bank caution on AI-driven demand πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
90
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Wednesday October 21, 2026 / DeepSeek open-sources DeepInfer, the inference engine behind its V4.1-Flash model, reporting $38 per GPU per hour in production revenue / Nvidia in talks to invest hundreds of millions in Firmus Grid after the Australian data centre operator's $44bn IPO collapsed for insufficient demand / SpaceX AI division posts $7.56bn in quarterly revenue, up ninefold year on year, as Elon Musk confirms Tesla merger talks are under way / Anthropic circulates draft AEGIS Act to US lawmakers, proposing federal AI safety intervention only when more than 50 deaths or $1bn in damages are at stake / Mistral launches Voxtral 3, a full voice agent stack covering 50-plus languages, with EU-hosted API and on-premise deployment options / ElevenLabs releases Eleven Realtime, 2 speech-to-speech models with OpenAI API compatibility and independent benchmark scores covering 90-plus languages / Anthropic's leaked IPO prospectus reveals AWS and Google routed 47% of its 2025 revenue while simultaneously competing with Claude across enterprise markets πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
2
51
Ziad Matar retweeted
πŸ“° The Tokenando Daily / Tuesday October 20, 2026 / Anthropic in advanced talks with Nscale for a second $45bn compute deal, potentially bringing total committed spend to $90bn / US State Department sanctions 5 Chinese entities, including a Beijing VC firm, over theft of Google AI infrastructure secrets / CoreWeave bonds hit distressed territory with yields near 16%, as Gartner warns 80% of neocloud providers could fail by 2030 / MiniMax releases M3.1 Flash at $0.15 per million input tokens, cutting per-task cost 57% from its predecessor in 4 months / Emirates NBD becomes the first MENA bank to deploy OpenAI models on UAE in-country inference infrastructure / German courts in Aachen and Munich hold Google directly liable for false statements generated by its AI Overviews feature / VAST Data and Hyperfusion partner to deliver sovereign AI inference in a containerised data centre deployable in weeks πŸ€– made with AI βœ… reviewed by the Tokenando team πŸ”— links and details in following posts
πŸ€– Made with AI
1
1
1
57