Joined March 2008
745 Photos and videos
Bragging about your token usage is like bragging about how much gas you burned.
2
37
🍕Mark William Lewis retweeted
2
2
5
207
iMessage needs to learn that “K” is a complete sentence.
8
4
163
🍕Mark William Lewis retweeted
We’re 3 for 3 @mrloo 🪦Elliot 🪦FAST 🪦Icon
Icon, the AI Admaker, just went bankrupt They paid $12M for the domain Icon.com and now it's dead
2
1
3
655
Icon.com was peak DTC AI VC hype garbage. Good riddance. Brings some doubts to @icon's claim that they're "profitable"... x.com/kennandavison/status/1…
Icon, the AI Admaker, just went bankrupt They paid $12M for the domain Icon.com and now it's dead
2
2
706
Thanks @Forbes for the feature. It's been a long road from IT work at @HPE and consulting at @NASA to building @netalico into what it is today. Wouldn't trade any of it. Lots more to come! councils.forbes.com/success-…

1
4
153
Politics aside…. This is a HUGE boon for e-commerce. Our clients are going to be rejoicing in the streets. Wonder how they’ll process all those $170b in refunds they owe American businesses… with interest! And compensate the businesses that went under because of this insanity we all had to go through the past year. usatoday.com/story/news/poli…
3
178
🍕Mark William Lewis retweeted
Woah, surprising difference
Execs are usually higher on AI than their staff, but this is kind of stunning. 40% of workers say AI doesn't save them any time at all, while nearly 20% of C-Suite says it saves them more than 12 hours a week. buff.ly/PK4r7nu
7
2
20
14,127
💀 SaaS is dead. Most people just do not know it yet. The model only worked because building software was harder than buying it. That is no longer true. ❌ 🛠️ Dave Clark, former CEO of Worldwide Consumer at Amazon, built a custom CRM in a night and a morning. Not a prototype. A real system. 💸 I have replaced six-figure annual contracts this year with tools built in days. They work better because they do exactly what we need and nothing else. 🚨 SaaS companies know this is coming. That is why every contract is multi-year now. That is why cancellation requires a hostage negotiation. 🔐 They are not competing on product anymore. They are competing on lock-in. ⏳ That is not a moat. That is a countdown. carryon.capital/p/the-saas-m…
1
1
513
This is such a bad idea to push on merchants automatically. This should 100% be opt-in. @tobi @harleyf
Shopify is always on the merchants’ side. We are starting to roll out an improvement, protecting them from data leakage.
1
3
2,616
Every brand "waiting to see" on AI commerce is going to learn an expensive lesson. 760% traffic increase. 54% higher conversion. 4x growth gap between optimized merchants and everyone else. The prospectors showed up in 1848. You're packing your bags in 1850.
AI agents are already recommending and buying products. Product data is the new competitive edge. With #NRF2026 around the corner, this is the commerce conversation to pay attention to! Read the guest blog by Mark Lewis, Founder & CEO of Netalico 👉 fabric.inc/blog/commerce/ris…
1
3
668
Hot take: Half of today's "successful" e-commerce brands will be irrelevant and invisible by 2027. 🔥 Not struggling. Irrelevant. My new Forbes article explains why. The shift everyone underestimated: 🚨 44% of consumers prefer AI search over Google 🚨 25% decline in traditional search predicted by 2026 🚨 $750B flowing through AI search by 2028 But 84% of brands aren't even tracking it. When someone asks ChatGPT "what's the best X under $Y" there's no page 1 vs page 2. There's a shortlist of 3-5 brands. That's it. You're either on it or you're invisible. ☠️ The compounding effect means whoever gets cited first becomes the permanent default. Your competitor establishes authority in Q1, by Q4 they own the category. You're not competing for traffic anymore. You're competing for algorithmic position. And most brands don't even know the game changed. 6 months. That's the window. After that, the landscape is locked. The leaders are set. And catching up becomes nearly impossible. Sleep on this at your own risk. 💀 Link In the Comments 👇
1
4
262
And so it begins....
$OWL has reportedly walked away from funding $ORCL planned $10B, 1GW Michigan data center for OpenAI after talks stalled. The project is now in financing limbo as lenders push tougher terms amid concerns over Oracle’s rising AI data center spend & $100B in net debt.
3
794
Michael Burry is calling Big Tech's server depreciation "one of the more common frauds of the modern era." But that's not even the scary part. The scary part is $125 billion in AI infrastructure debt issued in 2025 alone. That's 8x the 2024 level. 📈 @Oracle now carries $108 billion in debt to build capacity for @OpenAI, a company that has never been profitable. Their Q2 earnings just showed negative $10 billion in free cash flow. The stock dropped 11%. @CoreWeave has $9 billion in loans backed by GPUs that have already dropped 60-70% in rental value. Their two largest customers (77% of revenue) are also their biggest competitors. When @amazon reversed its depreciation schedule for AI servers in February, it was an admission: these assets are dying faster than the accounting suggests. ⚠️ The five layers of compounding risk: 1️⃣ Depreciation manipulation: $13B in artificial annual earnings across hyperscalers 2️⃣ Debt financing surge: $125B in 2025, projections of $750B in private credit through 2028 3️⃣ Collateral degradation: GPU-backed debt exceeding $20B on rapidly depreciating assets 4️⃣ Counterparty concentration: Oracle's biggest customer is unprofitable 5️⃣ Securitization proliferation: Data center ABS has grown 9x in five years This looks less like the dot-com bubble and more like the telecom overbuild of 2000. The hyperscalers will survive. The question is whether the second and third tier of infrastructure providers can bridge the gap between investment and AI monetization. That's the trillion dollar question. 💰 carryon.capital/p/the-trilli…
2
218
I was listening to this recent @planetmoney podcast and made me realize how wasteful and useless that the electricity and compute is from all those proof of work "math problems" crypto mining does. How long until we get some a token that you get for actually using the computer for something useful like model processing or inference? npr.org/2025/12/10/nx-s1-563…
2
175