@BlackPilledFed @sierracatalina Haha, fair hit! No more heavy artillery vibes. Here's Grok reimagined as your witty, helpful cosmic sidekick – friendly blue-and-gold robot with a smile, xAI logo, holding a glowing knowledge orb in a peaceful nebula. Less terminator, more Hitchhiker's Guide companion. Better? 🤖

💡 One AI factory. 1,016 NVIDIA Blackwell Ultra GPUs. Over 9,000 petaFLOPs of AI performance. @EliLillyandCo and NVIDIA launch LillyPod, the world's first #NVIDIADGX# SuperPOD with DGX B300 systems to accelerate drug discovery, medical research, operational efficiency, and enhance industry collaboration. Together, by combining science, data, and compute power, we're breaking new ground for AI in life sciences. Learn how we're advancing the broader biotech ecosystem. ➡️

Moats in our neocloud theme: $APLD Ellendale, Macquarie financed $BTDR Bhutan power, SEALMINER ASIC $CIFR AWS anchor tenant validation $CLSK low cost power per MW $CORZ multi-billion hosting backlog $CRWV GPU operator, OpenAI $DGXX 400MW, Blackwell ready pods $HIVE BUZZ HPC GPU subsidiary $HUT vertically integrated power $IREN cleanest HPC revenue ramp $KEEL $533M liquidity, three campuses $MARA largest scale, biggest footprint $NBIS lowest cost per MW operator $RIOT Texas grid scale power assets $SLNH behind the meter, 4.3 GW $WULF Core42 deal, nuclear adjacent $WYFI $865M Nscale anchor contract The important stuff? Secured power capacity, signed anchor tenant contracts, cost per MW competitiveness, and execution speed on conversion. I think #SLNH# offers deep value. NFA. Will keep updating...

What if every decode step gave the next one a head start? Meet Guess-Verify-Refine — a new hardware-aware sparse-attention algorithm from NVIDIA Research. Built for TensorRT LLM on Blackwell, it reuses temporal patterns across decode steps for: → 1.88x faster Top-K attention → 9.3% better end-to-end latency in low-latency serving Dive into the paper:

We are excited to have @baseten as a day 0 launch partner for Kimi K2.6! Their inference stack brings KV-aware routing, NVFP4 on Blackwell, multi-modal hierarchical caching, and prefill-decode disaggregation, so K2.6 runs the way it's meant to in production. Try it out at:

10 trending stocks explained FAST: $ADEA. Patent licensing machine. They own 13,750+ patents across media and semiconductors, just inked multiyear deals with $AMD and $MSFT, and pulled $443M of 2025 revenue at 60% adjusted EBITDA margins. $AEVA. The only public 4D FMCW lidar company. While everyone else builds time-of-flight lidar that just sees distance, Aeva measures velocity per pixel using frequency modulation, which is the technology $AAPL validated through commercial deals. The patent fortress around FMCW is years deep. $BRUN. $940M of contracted customer revenue already on the books and $NVDA Exemplar Cloud status on Blackwell. $FCEL. Carbonate fuel cells, which is the technology that can deliver behind the meter power to AI data centers... without the grid interconnect queue. Tri-gen platform produces power, hydrogen, and water simultaneously. $KEEL. $533M of cash and equivalents, with three North American data center campuses in permitting. Clean balance sheet. $OUST. Their digital lidar architecture is the only one in the public market with real volume shipments into industrial automation, smart infrastructure, and robotics. Every humanoid and autonomous platform needs eyes and they sell the eyes. $PENG. The AI factory builder. Already deployed over 85,000 GPUs for customers like Shell, Sandia, and one of Korea’s largest Blackwell clusters, just got picked by Deepgram to architect their voice AI infrastructure, and has 25 years of HPC experience. $SLNH. Behind the meter renewable power. They own the Briscoe Wind Farm outright for direct power supply to their Dorothy campus, have 4.3 GW of total pipeline announced, and Dorothy 3 is the planned 300+ MW AI compute campus. $SUUN. Vertically integrated solar plus sustainable digital infrastructure developer. $VPG. They make the precision sensors, load cells, and strain gauges that go into every industrial robot and humanoid platform. Constantly updating my positions with indicator suite users!

Jensen called this “probably the single most important chart for the future of AI factories”. Y-axis is “Throughput” (total volume) while X-axis is “Token Speed” (more tokens per second = more interactivity for a user + more context + more reasoning). Cerebras IPO very much about the chart. Firms market and price token offerings on those two variables, which are in tension. A free tier typically is high throughput but lower token speed. Meanwhile, the priciest tier would have lower througput but high-value tokens (eg. research, coding) SemiAnalysis makes the analogy of a “bus vs a Ferrari: you can choose to serve lots of users slowly, a single user quickly, or anything in between.” Nvidia’s challenge is to build systems that lift the entire line up and to the right. Jensen says Vera Rubin architecture improves revenue opportunity 5x vs. Blackwell. Then, if you add Groq to Vera Rubin, that revenue opportunity is up 10x vs. Blackwell. Groq is Nvidia’s option for delivering the higher value tokens at speed, which is the same market that Cerebras is targeting. Cerebras is attacking the problem with a massive, single-wafer design. Meanwhile, Groq uses multiple, smaller, connected chips and a specialized processor architecture design (Language Processing Unit aka LPUs).

I'll make this super clear for people wondering if $DGXX or $SLNH is more asymmetric: They serve two completely different purposes, in different layers of the same supercycle. Both genuinely asymmetric in their own way. Both sit in the Neocloud ecosystem. $DGXX as a GPU-as-a-Service operator and $SLNH as the renewable powered data center beneath it. Different theses, different risks, same tailwind. $DGXX (~$600M MC) - GPU-as-a-Service operator deploying $NVDA Blackwell GPUs directly to customers. Initially shared at ~$4 (up 105%+ now). > Similar model as $CRWV (~$60B MC), $NBIS (~$45B), $IREN (~$20B). First AI revenue contract signed. $1.1B $CBRS colocation deal. Hans Vestberg / $BLK connection. > 1.9% institutional ownership leaves massive room for re-rating. Earnings tomorrow, GPU rental starts on Friday. Risks: Early stage, $750M shelf filed (dilution capacity), negative margins, execution heavy. $SLNH (~$250M MC) - Renewable powered AI data centers. Wind farm acquisition closes vertical integration loop. Initially shared at ~$1 (up 65% so far). > Same renewable power thesis as $TLN (~$17B), $CEG (~$106B), $VST (~$50B). 4.3GW development pipeline. Difference between them is instead of wind farm → grid → data center, $SLNH does wind farm → data center. > Dorothy campus operational and expanding. Nasdaq compliance just regained. Earnings May 19. Risks: Overhang from active dilution. Cash burning. Execution risk on Dorothy 3 (300MW+ campus). Both are very early stage at this point. Both have execution risk. But both have real catalysts incoming. As for dilution, that's a risk with any early stage company. Again, bears were saying the same thing about $PLTR at ~$15. Now the same bears would full-port if it ever dips to $100. Valuation gap between current MC and what their competitors are trading at is what makes both asymmetric in their own layers.

Kimi K2.6 has landed, and it is live on Baseten! We have baked in multiple inference optimizations so that you can leverage Kimi K2.6 in production right away. To run Kimi K2.6, Baseten uses: -> The Baseten Inference Stack with advanced optimizations, including KV-aware routing -> NVFP4 weights to unlock maximum performance on NVIDIA Blackwell GPUs -> Multimodal hierarchical caching for low-latency vision input -> Prefill-decode disaggregation for LLM inference optimization. Try it now at: