Leading Edge Semi Shortages - a deep dive techinvestments.io/p/leading…

Investing in ASIC accelerator plays got you almost twice the return of $NVDA over the last 2.5 years (+290% vs +150%) basket: 40% AVGO, 40% MRVL, 20% ALAB

GPT Sol 6 is priced at a 50% discount to Claude Opus 5.5, basically the same as Sonnet. OpenAI is obviously trying to grab as much market share as possible. Of course it's working, I just already changed some of my API calls. Sol is a very good model.

Deutsche: "The DRAM supply-demand imbalance will be worse in 2027 and 2028" full summary: "Overall, we have increased our supply forecast while slightly tweaking our demand forecast downward. The net of these adjustments indicates that the DRAM supply-demand imbalance will be worse in 2027 and 2028 before nearing equilibrium by 2029 and potentially reaching oversupply in 2030 (depending on the pacing of fab buildouts). At the core of this outlook, our model still projects robust long-term demand well above historical levels. We see DRAM bit demand growing at a ~21% CAGR through 2030 (vs. historical trend in the ~mid-teens). Meanwhile, actual DRAM wafer supply is projected to grow at a ~+15% CAGR, above our prior est (~+12%) but still well below demand. As usual, our model distinguishes wafer supply growth from bit supply growth to better account for the trade ratio caused by HBM."

Why Google designed the TPU (recent talk by @JeffDean )

$NVDA is becoming a play on SpaceX

Broadcom CEO Hock Tan breaks down AI economics: Open-weight models burn $100B compute to make $30B in revenue. Frontier models spend $100B to make $120B in revenue. One of these won't be sustainable source: Goldman conference

While the market is overly focused on $AVGO market share in Google's TPU programs, the company is quietly becoming the silicon supplier for the dominant AI labs - which will have even bigger AI silicon demand than Google a number of years from now. Good take from Macquarie:

JP Morgan on Tesla Robotaxi post Cybercab launch event in Austin $TSLA "TSLA reported 1 mn cumulative miles of unsupervised Robotaxi operation with no human intervention. Cybercabs and the broader Robotaxi fleet are running early builds of FSD v15 for unsupervised operation, though these remain partial builds with ~40% of major improvement tracks merged. Customer-owned vehicles remain on v14.x. Ashok posted on X that the next v15 merge should enable more complete overnight/24/7 operations “next month or so.” Cybercab uses a ~48 kWh pack with 4680 cells and dry cathode, delivering efficiency of ~165 Wh/mi, unadjusted range of ~418 miles, and adjusted range of ~293 miles. The vehicle uses a single front motor of ~219 hp. Elon Musk posted on X that the Cybercab motor uses no rare-earth metals while maintaining the same range. Related details shared around the event included a drive unit that is 18% smaller, 25% lighter, and more efficient than other top-performing EV drive units. The drive unit features a simplified bar- wound stator and lubrication system and is designed for fully automated assembly with sub-10-second cycle times. The longer-term plan remains a multimillion- vehicle Cybercab network. TSLA cited the production line, engineering organization, field operations, and supporting infrastructure as the foundation for scaling the fleet. We size the global robotaxi TAM using a bottom-up, VMT-based framework spanning four geographies (US, Developed ex- US, China, EMs), each with distinct regulatory access curves, urban VMT shares, and competitive dynamics. The model layers five sequential cost phases, from current ride-hail pricing (~$2.00-3.00/mile) through to TSLA’s long-term Cybercab target ($0.30/mile), onto an accelerated VMT base that compounds faster as robotaxi availability expands (i.e., Jevons paradox). Robotaxi fleet size is built from paid miles, assuming ~65K annual miles per vehicle, a 200K-mile useful life, and ~30% deadhead miles, with TSLA monetizing owned fleet miles directly and owner- network miles through a ~20% platform take rate. This drives our combined TSLA robotaxi revenue projection to ~$300 bn by 2035 and ~$500 bn by 2040, with owned fleet economics still carrying the bulk of the dollar opportunity while the network layer adds more capital-light upside as supply scales. In this scenario, we estimate that the implied Cybercab fleet reaches ~35 mn vehicles by 2040."

Japanese humanoid robotics supply chain (Goldman)

Deutsche on the negative investor sentiment in Semis: Investor sentiment has soured materially from the enthusiasm seen in the first half of the year, with "peak cycle" fears common among many investors. Much of this bearishness stems from ongoing concerns of demand destruction caused by DRAM pricing and technological innovation to break through the "Memory wall". Per our meetings and strong earnings reports last week, it's clear demand for AI compute is not slowing, with investors now assuming 2027 hyperscaler capex will reach $1.2–$1.5t, up from ~$800b in 2026. In addition, demand has broadened beyond hyperscalers alone, with neoclouds, enterprises, sovereign AI initiatives, and emerging AI labs all ramping spending and becoming an increasingly important part of the demand landscape. As such, demand is not the limiting factor; rather, memory and, to some extent, power are the primary bottlenecks. The memory wall and ongoing supply constraints are leading customers to rethink and optimize future product roadmaps, with much investor attention focused on potential HBM de-speccing of next-gen processors, as well as the impact of higher DRAM pricing on total AI spend. In turn, this issue has also spurned competitive innovation on the inference side of things, resulting in a higher number of emerging solutions such as SRAM-heavy inference chips, custom HBM base dies, and many variations on the HBM theme (HBF, etc). This situation appears to remain highly fluid, with engineering conversations ongoing and visibility within the supply chain very cloudy. Nevertheless, most management teams we spoke with believe the issue to be one of economics rather than technology, with aggregate DRAM demand likely to remain high.

Apparently customers aren't vibe coding $SNOW Easy money

$MSFT new reporting lines are smart marketing by Satya

Rather than looking at all the mainstream benchmarks - which every AI model is optimizing for anyways in training - out-of-sample/ real-world workloads are far more useful for evaluating the various LLMs This is a real, very complicated test designed by @dhh, from the @lexfridman podcast. The workload was basically to turn a Python library into a Rust one. Fable designed the plan and Opus completed the job as he ran out of tokens. So Claude completed the job in 45 min overall. Then, he gave the plan from Fable to a variety of other top models to see if they complete the job: Even with Fable's plan, it took Deepseek Pro four times as long to complete. Kimi K3 took forever. Sol needed twice as long as Claude. Also Grok 4.6 could complete it. Despite all the hype on X on open-source models, they are still far behind. Time is money - I'd rather pay Fable to do it fast than wait all day for Deepseek and Kimi

Our buy thesis on $AVGO techinvestments.io/p/broadco…

Interesting call on the Google - $RDDT partnership with a former lead at Google, as well as why AI talent is leaving $GOOGL The read is that Reddit's negotiating position has deteriorated materially His framing is that the original agreement was a function of where large language models sat at that specific moment. Bard was becoming Gemini, hallucination rates were high, and the model had no grounding in current events. Training produces a model that is effectively six months stale by the time it ships. Reddit solved a narrow, acute problem: real-time human commentary at breadth and depth. The competitive set for that data was thin. X, Meta's properties and Threads are walled gardens aligned with rival frontier labs and were never gettable. Reddit was the one large corpus of unfiltered human language actually available for purchase, which is why both Google and OpenAI ended up there. The price reflects how little this mattered to Google's P&L. Roughly $60m: "For Google, $60 million to buy specific data is not a lot of money." His view is that the cash was the least interesting component. The valuable consideration was prompt data flowing back—what the user asked, and what they asked that led to a click through to Reddit. Both sides of the intent coin. He connects this to the trajectory of Reddit's advertising business, which has scaled well beyond what a new sales team and new infrastructure alone would explain. On the widely discussed point that AI Overviews traffic converts poorly, he broadly accepts it but argues the second-order effect dominates. Reddit held primacy in the citation slot, so volume was high even if quality was low, and the learning from that volume compounded. Google has since connected its search index and corpora more directly to the model layer, so the original grounding gap has largely closed. YouTube citation share has overtaken Reddit. Google News, Merchant Center and Places cover most of what Reddit was a shortcut to. His read on Reddit publicly signaling it might walk is that this is negotiation conducted through the press, and that it implies Google came back with worse terms—likely stripping preferences and the prompt data return rather than simply cutting the number. Google's standard posture on data rights is full and unrestricted use, and carve-outs on usage are not how Google contracts. "They probably don't need it. They probably want it." He expects a deal—the relationship is warm and mutually beneficial—but on terms that reset Reddit's expectations. Money is the secondary variable. The variable that matters to $RDDT holders is whether prompt-level signal keeps flowing. Asked whether Google would simply take the data if talks collapse, he says no on cultural grounds, that it is not in the corporate DNA to do that after the fact. More interesting is his description of how frontier labs behave when sued over training data: they do not settle, because a settlement establishes a price and invites every other rights holder. They litigate, spend, and drag it to a quiet resolution specifically to avoid setting precedent. Independent of the Google relationship, he identifies the harder issue: Reddit sits in the middle of a considered purchase journey with nothing to sell at the end of it. A user researches a bike on Reddit and buys it somewhere else. Two steps, and the second one is where the margin lives. The old funnel involved ten websites and fifty data points before purchase. That discovery phase is now collapsing into the chat interface, and the losers are the intermediaries that monetized the journey rather than the transaction. This is a Gemini problem, a Claude problem and an OpenAI problem simultaneously, not a Google-specific one. Search advertising worked because the system was deterministic—a tree you navigate from trunk to branch to leaf, ending in a transaction. Token predictors give wildly different answers to marginally different prompts, and the labs do not fully understand their own models' behavior post-training. Tuning that for advertiser ROAS is closer to dark arts than to keyword auction mechanics. The deeper constraint is grounding. Merchant Center is the largest product data repository in the world, Places the largest inventory of shops and locations, and every flight and hotel has been tuned within an inch of its life because advertisers were trained over two decades to feed that system. OpenAI has none of it. He is measured on the disruption question rather than dismissive. He cites Walmart attributing roughly 20% of traffic to OpenAI as evidence the relay is real, and he sees a credible alternative path: merchants exposing their own data through open interfaces that models come and fetch, rather than piping it into Merchant Center. That inverts the current architecture, but it needs an ROI case to bootstrap and there is a chicken-and-egg problem. His conclusion is share erosion and ad revenue siphoning, not collapse. On the suggestion that OpenAI should just acquire an ad tech engine, he is dismissive for the right reason: buying keyword-era infrastructure is buying an internal combustion engine in an electric vehicle world. A paragraph-long voice prompt carries vastly more intent than a three-word query, and the extraction method has to be built for that, not retrofitted. "Google has innovator's dilemma on steroids." A USD 250bn high-margin advertising business, powered by data that advertisers were trained to supply, cannot be hard-switched to Gemini. The chosen path is to infiltrate Search with AI and tolerate a messy middle until ROAS re-stabilizes on the new medium. He is candid that the current state is worse than Search at its peak, and he references the reported history of deliberately degrading search ad quality to increase monetization as evidence that the profit motive is explicit. On Google execution speed - Every objective must ladder from the most junior contributor up through director and VP. Search sits at the top of the tree, Android just behind, peripheral products have no influence. Strategy gets negotiated between silos, which is slow. "innovation is by definition outside of that data structure." Work outside the ladder is unsanctioned, and unsanctioned work costs you promotions and raises. He extends this to DeepMind, which he believes is now materially less isolated than it was and is being pulled into commercial delivery, and offers that as the explanation for researcher attrition to Anthropic and OpenAI. Not compensation—loss of research autonomy.

Good call on Bloom Energy $BE with a director on Crusoe's energy team TLDR: Constructive on $BE through the speed-to-power window: everything they can build gets sold, the regulatory-constraint use case is expanding, and the edge market is a genuine second leg. The bear case is not execution risk, it is terminal value—an unsubsidized, unconstrained power market where Bloom's cost curve has not moved enough. Key insights: Behind-the-meter adoption is not a technology preference, it is a hedge against binary regulatory risk. Moratoriums, temporary pauses and Governor Abbott's ERCOT announcements do not degrade a project's economics—they kill it outright, and approved power becomes worthless if a data center moratorium lands on top of it. He has seen two separate tenants elect to proceed with behind-the-meter projects alongside their grid portfolio, explicitly as diversification against that risk. Crusoe deliberately sites behind-the-meter gas away from rural communities to strip out both grid-connected regulatory exposure and community sentiment risk. Texas is the tell. It has historically been the easiest place in the US to get grid power because ERCOT is deregulated—no capacity market booking, no requirement to point specific generation at a specific load, with scarcity pricing left to incentivize the build-out. The fact that delays and regulatory uncertainty are showing up there is what has shifted tenant behavior. Working from a 100 GW / five-year data center demand frame, which he treats as aggressive but attainable: Six months ago: roughly 70 GW served by grid, assuming turbine production picks up, no fuel constraints, and several other unlocks. Some observers penciled 10-20 GW of SMR by end of period. Today: near-term grid share compresses to perhaps 50-60 GW, with the forward split moving closer to 50/50. Then it reverts. Once ratepayer protection is formalized, the grid reasserts as the cheapest and most reliable long-term supply. The mechanism for reversion is the Arizona construct APS has pushed—growth pays for growth, where all incremental upgrade costs are borne by the data center operator. He expects that to be cemented and formalized across every market, and expects it to take six to twelve months before politicians stop feeling their seats are threatened. The Bloom Bull Case He does not dispute near-term demand at all. "as much reliable Bloom capacity or solid oxide fuel cell capacity that can come online, will be deployed." If US deployable manufacturing produces 2-8 GW over the next two years, it gets absorbed. The premium is not a problem for buyers whose binding constraint is capacity. He also believes the product works. Hundreds of megawatts is deliverable, the systems are reliable, downtime is low because units are swappable, and the architecture is fully modular. The Bear Case Is About Price - "I don't see how they compete on price." Ten years out, in an unconstrained power market, he does not believe Bloom is competitive. He specifically does not believe the roughly 10% annual cost-down, on the grounds that the technology is structurally hard to make cheaper. He also flags a cost item he thinks the market underweights: the full system swap over a ten-year cycle, which makes ongoing O&M more expensive than a gas gen equivalent. Initial capex is the wrong lens; actual LCOE is critical to assess. He extends this into a coherent explanation of Bloom's own behavior. On the question of why they have not simply built the next facility if demand is as strong as claimed — his answer is not that they are sandbagging. He thinks they are building as fast as they can, and that the constraint is the supply chain, not the building. Solid oxide is an extremely small US market. You can put up a structure; scaling the full assembly chain behind it on the same timeline is the harder problem, unless more of it moves offshore. Crusoe Has Zero Bloom Projects Today - "Today, actually, we don't have any projects that are relying on Bloom fuel cells." The strategic rationale for staying at arm's length: "We've been a follower in this instance... we will accept it once the utility does." Buying 500 MW of Bloom directly means absorbing the regulatory risk that utilities may not accept solid oxide as high-rated ELCC capacity. Let the utility carry that. Turbines, recips, aeros and engines are all accepted technologies with known extreme-weather behavior and established effective load carrying capacity ratings. Solar, wind and battery now have them too. Solid oxide does not, because utilities have not yet observed large-scale fuel cell fleets through heat events and cold snaps. AEP has gotten comfortable off the back of the roughly 80 MW deployment plus smaller installations and Bloom's published test results. That is the template, and it is why the next few hundred megawatts of live operating hours matter far more than any order announcement. Where Bloom Actually Wins His model is not that Bloom wins on merit in a fair fight. It is that Bloom is the path of least resistance when a specific constraint blocks a project that already has hundreds of millions of development dollars sunk into it. If the binding constraint is price, Bloom loses. If it is speed to power, Bloom sometimes wins. If it is emissions, air permitting, noise or a regulatory restriction someone failed to plan for, Bloom wins. He reads the Nebius Vineland switch from gas gensets to solid oxide exactly this way — anti-genset pushback threatening a contract worth billions, with an obvious substitution available. He expects more of that, in lumpy project-specific chunks rather than as a smooth share gain. He also identifies the next leg of NIMBY-ism, which he thinks is underpriced: if communities dislike data centers, they dislike new gas generators considerably more. A fuel cell is lower emissions, quieter and a different class of asset. He calls it artful. That is a real, non-obvious tailwind. The Edge and Inference Market Is the Bigger Prize Crusoe is planning heavily for 10-50 MW modular builds, which he sizes at 20-40 GW over five years and would anchor at 20 GW in isolation. Amazon, NVIDIA, Tesla and xAI are all chasing the same edge market. "Bloom will leapfrog any gas combustion." In metro locations you cannot air-permit gas gen at all — this is a hard prohibition, not a noise preference. But he immediately caps the enthusiasm: much of that edge capacity is low-hanging fruit, converted Bitcoin sites at 5-18 MW with existing grid interconnects. "Grid power will always beat it." Bloom is the answer for incremental capacity and backup, not the base case. This is why he holds solid oxide at roughly 10 GW of the 100 GW mix, against 60-70% backstopped by combustion gas. He validates the native DC output argument, but for a better reason than efficiency. Fewer conversion losses lower the price, yes. The larger point is that it removes dependence on transformers and switchgear — equipment that is not only expensive but carries lead times that are themselves the binding constraint. A 345 kV breaker is two years out. Engineering around that bottleneck is precisely the kind of scenario where a project with sunk capital selects Bloom. What Changes His Mind He names two variables explicitly. Power price curves — if Bloom's costs do not fall, or if turbine pricing keeps rising, the relative position shifts and Bloom is in the money reasonably soon. And political sentiment, which he calls a big unknown and which he thinks is the more likely driver of fuel cell share than any technical milestone. Notably, a successful 200 MW deployment alone does not change his view. He already believes they can do it. source: Tegus

Goldman: HBM ASPs to increase 90-100% yoy in 2027

DRAM shortages are increasing - $MU not able to meet half of data center demand

I halved my $GOOGL position: - They haven't released a new model in ages - Gemini Pro regularly crashes - A lot of the best people in AI just left - Search is 60% of profits, which is going to get largely disrupted - The parts I like are GCP and Youtube, but these are small parts of the business (35% or so) - 27x forward PE, fairly expensive given the risk in Search I've kept half for the moment on the hope they'll release a cool model soon, but all of this position is pure profit. I first invested in Google in 2011 and have gradually been selling over the past two years as Anthropic is just on a much faster innovation cadence and given my bearishness on Search. Anthropic is launching one impressive model after another. I do all my coding with Claude. Whenever I try one of these open-source models, they're garbage. It's called 'benchmaxing', as Palantir explained well - i.e. make the model look good on predefined benchmarks, but not being able to handle real workloads that engineers need. My impression is that Anthropic is running away with this and that the gap with competition will only increase from here. Anthropic has the cash to build custom data sets, has the talent, and the compute. Google seems to be lacking in execution. I don't want to write them off fully, but it certainly doesn't feel like high conviction.