William Peets (@WillPeets) header
William Peets (@WillPeets) profile picture

President and CRO Energy Group Capital

San Francisco
Joined December 2009
Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)Photo by William Peets (@WillPeets)
This is the defining concept of our time nitter.cf/CarlBMenger/status/197…
„Don’t worry Bro, we got them magic money printer.“
1
13
3,207
William Peets retweeted
24hrs ago: Openai can do the math to break cryptography and send all of your cryptocurrency tokens to 0 Now: Openai can't do the math on their own revenue
OPENAI RECENTLY TOLD INVESTORS ITS REVENUES WERE APPROACHING $50BN ON AN ANNUALISED BASIS AT END OF SEPTEMBER, FAR SHORT OF THE $70BN REPORTED - FT
42
74
7
890
43,476
William Peets retweeted
lots of things to unpack: 1. our largest deal yet (big size) 2. our 1st investment-grade offtake deal (great credit) 3. a new product offering to help scale our business - that k3 annc was a part of it you think you know what USDai is, but the design is improving each month
Largest loan in the book yet. $128.9M for 32 NVIDIA GB200 NVL72 (2,304 GPUs) in BC, Canada. Backed by an investment-grade offtaker: a high-credit-quality customer contracted to purchase the compute capacity. 6.5% in escrow. 9% once funded. 3-year term.
11
12
139
16,418
William Peets retweeted
Compute is rapidly becoming one of the largest credit markets. @Bullish (NYSE: BLSH) is entering with a $100M stablecoin liquidity facility for USDAI. The facility will support USDAI loans backed by the AI buildout.
18
26
12
185
40,274
William Peets retweeted
young americans protesting datacenters forget we offshored -uranium enrichment in the 90s and spent 30 years buying fuel from Russia -same story in rare earths and China controls 90% of magnet production, making it impossible to build robotics without Chinese parts -same with pharma precursors, all your drugs are from China if you don’t want compute built here i hope you like renting it from countries that want to win
79
227
36
2,026
130,446
William Peets retweeted
total raised: $7.5b total 24h revenue: $1,306 15,418 years to return investments
533
575
312
4,164
725,006
William Peets retweeted
I've been spending a lot of time thinking about and analyzing why the AI trades have been selling off. The easy answer is to say "A lot of things" and move on, but I think there is one major proximate cause that I hadn't considered. In the first chart below you can see that the Nasdaq topped the day of Warsh's first FOMC meeting. I had noticed this previously, but I wasn't sure how Warsh's previous meeting would flow through so directly to the AI trade. In the second chart you will see "real rates" measured by the 10y treasury - the 10y inflation breakeven number. Long term inflation expectations haven't moved up much, but treasury rates have. This leads to a tightening of financial conditions, and in this case ~50bps higher real rates. And Warsh is signaling more hawkish measures ahead. The committee has begun to indicate hikes (even as long term inflation breakevens haven't moved), and, perhaps more critically, Warsh is an ardent balance sheet hawk. What this means is that Warsh plans to shrink the balance sheet, probably by letting things like MBS and long duration treasuries roll off over time. Removing a buyer/holder of longer duration assets from the market also has the effect of raising real rates and tightening financial conditions. This all comes at a time when the hyperscalers are doing more capex than ever, and the entire AI complex is dependent on that capex for their own EPS targets. To fund this capex the hyperscalers are issuing more debt than ever by wide margins (see chart 3). This debt needs to be absorbed amongst tighter financial conditions more broadly, and potentially at the same time as Warsh is plotting to remove one of the major holders of duration from the market. Of course the hyperscaler fundraising is miniscule in comparison to treasury issuance, but the increased supply of treasuries raises their cost of capital. The AI buildout is also one of the longest duration trades in the market. So real rates rising has a big impact on the value of the future cash flows from the data centers being built. This would probably all be fine if the hyperscalers were rolling in free cash flow like they once were (and thus didn't need the debt to fuel the buildout), but they've already largely tapped that (chart 4). And thus the market is punishing them for this spend rather than rewarding it. AMZN/GOOG are flat on the year and META/MSFT are down 10% and 20% respectively, even as NDX is +10%. All of this has the net effect of constricting the AI trade. All of this added pressure on the hyperscalers leads to lower capex expectations farther out, which flows through directly to EPS expects. While analysts haven't begun to move semis EPS expects down yet, it's possible that the market is sniffing out this dynamic before the analysts. None of this is helped by all the AI-specific news like efficiency breakthroughs, Chinese hardware entrants, etc. And if I'm right about this then the worst part of it is that it's actually surprisingly difficult to trade on from here. Hyperscalers might begin to get rewarded if they cut capex, but that's difficult for them to do when they mostly seem to feel that AI is existential to their business. Semis have verifiably high backlogs, but they remain vulnerable to this whole dynamic. With Warsh this week maybe we will get some color and clarity, but I doubt it. He's been very clear about wanting to remove forward guidance from his mandate. I'm not a great rates trader, but maybe the opportunity is somewhere in there. Have a good trade idea for this dynamic? Let me know.
40
34
7
409
55,433
William Peets retweeted
Closed-door: hyperscalers and large credit funds are financing the biggest AI compute buildouts. Accessible and tradable onchain via USDAI: GPU-backed loans for the compute middle market, a market with up to $2T of capital demand. @_ConorMoore on @FullyVested_Pod with @patfscott and @defidave
6
6
42
4,473
William Peets retweeted
USDAI breaks $400M TVL again. sUSDai APY remains steady above 8%, the highest among yielding stablecoins with $100M+ TVL. This time, traction is driven by the $10 trillion demand from the AI infra buildout, turning demand for GPU financing into onchain yield.
9
16
1
84
13,350
William Peets retweeted
$10 trillion+ of AI hardware spend needs financing. > AI clouds buy GPUs > AI companies pay for compute > Capital (sUSDai) earns a share of the spread between contract value and build cost USDAI's yield comes from that massive demand. @0xZergs on @stabledash
6
11
2
62
7,971
William Peets retweeted
According to @stablewatchHQ data, base:0x0b2b2b2076d95dda7817e785989fe353fe955ef9 achieved 7.76% APY over the last 30 days. Yield Paid Out (YPO) to holders is closing in on $2m a month.
There’s a clear pattern showing up in the top stablecoin yield opportunities on @StablewatchHQ. Once you filter for $10M+ in TVL, this is what you get ↓ @UnitasLabs $sUSDu – 15.72% APY Yield comes from a basket of market-neutral strategies, including funding-rate capture, lending, trading fee accrual, and JLP yield capture. @Main_St_Finance $msY – 12% APY Yield comes from a delta-neutral CME options strategy, with part of the yield directed to an insurance fund. @noon_capital $sUSN – 10.76% APY Yield comes from capital deployed across delta-neutral strategies and RWA-backed assets. @PikuFinance $USP – 10.09% APY Yield comes from USD stables being deployed into DeFi strategies and yield-generating onchain/offchain assets. @gaib_ai $sAID – 9.37% APY Yield comes from RWA exposure tied to treasuries and AI infrastructure financing, backed by borrower interest payments. @avantprotocol $savUSD – 7.83% APY Yield comes from avUSD reserves being deployed across chains and protocols through managed strategies. @infiniFi $siUSD – 7.63% APY Yield comes from various onchain DeFi operations. @YuzuMoneyX $syzUSD – 7.50% APY Yield comes from yzUSD being deployed into onchain DeFi strategies, with protection layers mentioned by Stablewatch. @USDai_Official $sUSDai – 6.92% APY Yield comes from treasuries and AI infrastructure financing, with borrower interest payments feeding the strategy. @Neutrl $sNUSD – 6.13% APY Yield comes from liquid crypto-native stablecoins, yield-bearing assets, delta-neutral strategies, and hedged OTC positions.
13
6
1
47
6,839
William Peets retweeted
June was a big month for USDAI across loans, yield, liquidity, and APAC expansion. $98.1M largest loan to date $20M cumulative yield crossed $100M sUSDai liquidity facility with @0xfluid $300M APAC compute financing JV with BSQ
10
11
3
57
8,451
William Peets retweeted
Replying to @DavidSacks
More likely story: The "jailbreak" wasn't super serious (a situation anyone who has ever received bug reports is familiar with), Anthropic thought the demand to halt model was absurd, and Fed Gov used opportunity to punish and humiliate Anthropic for the prior sins of not bending knee. Anthropic has more credibility on such topics than Washington
150
137
20
3,110
160,566
William Peets retweeted
History's first trillionaire is a guy who catches rockets out of the sky with chopsticks and beams internet to every dead zone on the planet. Same guy ships cars that drive themselves, humanoid robots for the factory floor, brain chips that let paralyzed people move a cursor with pure thought, and an AI running on a supercomputer his team stood up in months instead of years. And the people crashing out about his net worth are doing it on the app he owns. The same app governments spent years trying to censor. You cannot legislate a rocket into orbit.
1,857
11,209
633
69,701
2,462,526
William Peets retweeted
3/1: "four to five weeks" 3/9: "very soon" 3/16: "won't be long" 3/23: "very good and productive conversations" 3/26: talks to end the war are “going very well" 3/29: "I think we'll make a deal with them, pretty sure" 4/1: “very shortly” 4/6: "They’ve made a proposal, and it’s a significant proposal." 4/8: ""A big day for World Peace!" 5/18: "we’ve had very big discussions with Iran" 5/23: "will be announced shortly." 6/1: "rapid pace" 6/11: "next few days"
53
323
27
1,468
129,828
William Peets retweeted
Citadel: ''Unlike traditional software, where the marginal cost of serving an additional user is close to zero, AI carries a meaningful and ongoing compute cost. The economics therefore depend not simply on what the technology can do, but on whether the productivity generated is sufficient to justify the resources required to deliver it.''
17
81
19
495
48,391
This
Price wars are always great for equity valuations 😭😭😭😭😭😭
1
1
120
William Peets retweeted
I’m in the process of revoking permissions rn. Better to be safe than sorry, especially if mythos is imminent.
I usually think we overdramatize the shit out of everything in this space. Every other week something is "the end of crypto" and then life just goes on as usual. But I just had a very long conversation with the incredibly talented white hat hacker @MevenRekt about Anthropic's "Mythos" (its public version will likely be called Fable 5) model and honestly, this one hit different. When some of the most talented white hat hackers and security experts are genuinely concerned about DeFi and crypto security and hence are taking every possible precaution to protect themselves, how could we not be? Quick recap for everyone who missed it: Anthropic built an AI model so ridiculously powerful at finding software vulnerabilities that they kept it locked away, only accessible to 150 handpicked organizations like Google, Microsoft and JPMorgan through a program called Glasswing. That thing has already found over 10,000 critical vulnerabilities across the world's most important software. The public light version of it, not even the full Mythos, found a critical flaw in an established protocol like Zcash within 24 hours. A flaw that had survived four years of scrutiny by some of the best cryptographers on the planet. Now they're about to release Mythos / Fable 5, a first restricted public version with similar capabilities. Think about what that means. The cost and skill required to find exploitable flaws in smart contracts is about to drop to basically zero. For DeFi this should be a massive wake up call. Unaudited protocols will become sitting ducks. Known exploits will get replayed on forks around the clock. Even small projects will get targeted simply because trying costs next to nothing now. Look, I'm not trying to spread panic here and I might be overdramatizing it. But I'm also not going to pretend this isn't serious, so I'm simply trying to wake up as many of you as I can to mitigate the potential damage as much as possible. I spent the last few hours revoking token approvals, moving funds to clean hardware wallets and cutting exposure to protocols I don't fully trust. It is a shit ton of effort and incredibly annoying, but I'd rather feel stupid for being too cautious than lose funds because I was too lazy. It might not be as acute as some people will make it sound. But the direction is crystal clear. If you haven't locked down your security yet, do it today. Not tomorrow, not next week. Today. There is nothing to lose by taking the right precautions now, but a lot to lose in the future if you haven't. Every help in sharing this for max awareness would be appreciated. Don't be the low hanging fruit hackers think you are.
7
1
33
9,921
William Peets retweeted
Lighthouse, the USDAI 2026 YTD report, is live. $44.6M protocol ARR. $13B borrower pipeline. $98.1M largest loan. 7.0% average sUSDai YTD yield.
29
46
9
280
82,161
William Peets retweeted
Replying to @aeberman12
Systemic buffers have worked thus far, but are being steadily ground down. Buffers eventually exhaust. Can kicking is not a solution. This chart clearly illustrates the inverse relationship between between commercial crude inventories and price.
1
5
284