Skip to content

Pulse - May 16

16 May 202622 min
ContentsOverview

The market is paying two bills at once. AI wants cheap power, cheap money, and infinite patience; inflation is charging for all three upfront. Equities can still hide behind earnings. Crypto has less cover: it needs flows, collateral, and catalysts that arrive before traders lose interest. Overall, narrow leadership, and little room for disappointment.

Key Takeaways

  • Retail sales rose, but inflation did part of the work.
  • Consumers feel recessionary while the S&P index trades records.
  • CPI was +0.6% m/m, +3.8% y/y; core CPI was +0.4% m/m, +2.8% y/y. PPI final demand was +1.4% m/m, +6.0% y/y. Both above expectations.
  • Equities still pay for AI capex and earnings; crypto pays for flows, liquidity, and catalysts.
  • AI has the bubble symptoms, hyperscaler capex estimates through 2030 have risen about 25% since October, and AI infrastructure spending can push inflation higher before productivity pulls it lower.
  • AI is profitable but capex-heavy; crypto is liquid but catalyst-sensitive; macro just removed the Fed-help crutch.
  • BTC sets tempo; if the thesis needs lower oil, lower rates, and perfect positioning, size it smaller.

Macro: No Cut To Lean On

Zooming out, two key variables kept confronting each other. On one side, the Middle East energy risk keeps inflation pressure alive. On the other, AI raises the first credible path to growth above the long 1.9% U.S. per-capita trend. AI builders are spending upfront while revenues arrive later, and governments are already carrying heavy debt. At the same time, energy accounted for more than 40% of the monthly CPI increase.

The main risk was a hot CPI/PPI week forcing the market to price no Fed help while longs are already crowded. CPI came in higher than expectations (+3.8% YoY vs +3.7% estimated; +0.6% MoM); PPI rose 6%, also above expectations and matching 2022 levels.

Growth sentiment is weak and consumers are angry, but inflation is not giving the Fed enough reasons to cut easily. Consumer sentiment has been sending a warning that equities continue to ignore. University of Michigan sentiment fell to 48.2 in preliminary May data from 49.8 in April, with current conditions down 9% month over month while Fox polling showed 70% of voters saying the economy feels like it is getting worse.

Labor's share of U.S. national income has fallen to 53.8%, the lowest since records began. Meanwhile, AI productivity gains are more likely to accrue to companies and shareholders unless labor gets bargaining power back. That is bullish capital, bearish social stability, and dangerous for broad consumer demand. Asset owners are still getting paid while wage earners are absorbing the inflation, energy, housing, and job-security shock.

Despite stocks at record-highs, consumer sentiment in the U.S. is sitting at outright crisis levels. The University of Michigan index dropped again in May to 48.2, a new all‑time low, with current conditions at 47.8 and households’ own financial situation back to 2009 stress points. That’s 13% below the 2008 GFC trough and even weaker than the 1980s recession bottom. The economy still prints okay headline data, but the median voter feels worse than during prior crises.

This is a bad mix: cuts get pushed out, real rates stay restrictive, and the market has to justify multiples without the “Fed put” doing the work.

AI: Crowded, Profitable, Fragile

AI infrastructure spending can push inflation higher before productivity pulls it lower. That is the flaw in the “AI solves everything now” thesis. Markets are treating AI like a duration asset with earnings. That makes sense until rates stop cooperating.

Alphabet is now pressing Nvidia for the largest-company crown, cloud growth is still strong, and semis have added trillions in market value over a short window. The businesses are solid; the positioning around them is crowded.

The bubble checklist is visible: euphoric posting, infinite payback horizons, and arguments that the move can only be early. The better defense for this cycle is cash flow. Hyperscalers earn money, buy chips with balance-sheet strength, and sell products into real demand. The weakness is capex reflexivity. If every model assumes more compute, more spend, and higher future productivity at the same time,

the trade has little room for disappointment.

The infrastructure buildout arrives before the productivity dividend. Power demand, data centers, chips, and financing costs hit first. Efficiency gains come later and with uncertain timing. That means AI can be pro-growth and inflationary in the same window.

Crypto: BTC Holds The Clock

BTC — First slipped under $80k on bad macro news. Then it snapped back quickly. Structural buyers exist, but fast money pukes when inflation, oil, and rates move against it. The CLARITY Act, which draws more clear lines between SEC and CFTC jurisdiction, helped lift spirits pushing price back above $82k temporarily.

ETH — Trading near $2.30k it has enough support to avoid irrelevance, but not enough strength to lead. Product flows improved with $77.1M in weekly inflows, and the Ethereum Foundation’s May 11 update shows steady protocol work around Hegotá, FOCIL, and native AA, but traders are not paying for engineering cadence. ETH needs $2,400–2,450 to wake up DeFi beta. Below $2,230, the market treats the ETF bid as dead money.

SOL — Still the best major for selective rotation. The setup works while $90 holds and $100 breaks with spot volume. Better than ETH for momentum; still hostage to BTC risk.

HYPE — Best liquid expression of real crypto revenue, HIP-3 OI has crossed $1.4B with tokenized stocks and commodities among the active markets, and HIP-4 as fresh catalyst (1%, 10%, 30% and 50% of Polymarket’s volume, HIP-4 would add about $347K, $3.5M, $10.4M and $17.4M per year respectively) with USDH collateral as the force multiplier. That’s exactly where this week’s big news hit, as Coinbase announced the acquisition of Native Markets’ USDH to increase USDC’s kingmaker status. USDC already sits near ~$5B on Hyperliquid, 2x YoY. With Coinbase becoming the official USDC deployer, the exchange will be sharing about 90% of reserve yields, which is on the order of ~ $150 million yearly from $5B in USDC. Circle handles technical USDC operations, and both stake 500,000 HYPE tokens to launch it.

ZEC — Privacy has momentum, shielded supply is at a record zone, and the quantum roadmap gives the thesis a fresh technical angle. The thesis is now understandable to anyone: Bitcoin got institutionalized; privacy became scarce; Zcash offers a cleaner regulatory/custody path than Monero. Still, vertical charts punish late entries.

AAVE — Still in damage control mode. On May 7, Aave liquidated the remaining rsETH positions tied to the Kelp DAO attacker on Ethereum and Arbitrum, pushing the DeFi United recovery closer while the broader rsETH/legal overhang remains unresolved. Days later Aave v4 reached $50M in deposits. Although buybacks remain paused, the longer-term outlook is still good. V4’s reinvestment module can put idle liquidity into approved low-risk strategies, with simulations lifting stablecoin APY from 4.00% to 4.93%. For buyers, conviction certainly increases after even more recovery clarity or buyback cadence returns.

PENDLE — With $300M+ of TVL linked to STRC, MicroStrategy’s credit is doing the legwork. STRC is Strategy’s Nasdaq-listed perpetual (not collateralized by Strategy’s BTC holdings), paying 11.50% variable annualized cash dividend on a $100 stated amount, with the rate adjusted monthly. Saturn brings that dividend stream onchain through USDat / sUSDat, and Pendle turns it into PT/YT markets: fixed yield of up to 18% APY, directional yield bets, and LP carry. This is quite bullish long-term, proving demand for yield-bearing assets moving freely through DeFi as collateral, leverage, duration, liquidity, and basis.

SPK — One of the largest beneficiaries in terms of ETH inflows after the rsETH hack which negatively impacted Aave. Spark being an Aave v3 fork, but where efficiency-mode looping is only enabled on the ETH-wstETH pair), has doubled its ETH deposits in the last month, surpassing the 500k mark. This comes right after also benefiting from $500M+ in stables when the rsETH incident happened. SPK went up >80% and Spark TVL moved from roughly $3.8B to above $5B, as more than $1B fled into the venue in a matter of days.

Spark is one of the few DeFi names showing operating resilience while lending conditions weaken. Q1 gross protocol returns were $31.5M, net protocol returns $6.91M, net surplus $3.46M, treasury $46.1M, and SPK buybacks hit $986k. Distribution rewards became the largest net-return contributor as lending spreads compressed.

TON — With Telegram and Durov’s leadership now in charge, distribution plays a major role. Telegram’s May 7 update added guest AI bots, bot-to-bot interaction, profile-connected automation, and Cocoon-powered AI search for stickers and emoji.

MEGA — While it builds its ecosystem, MegaETH is making USDm the internal economic engine: apps denominate in USDm, USDm rewards flow to the Foundation, and those rewards buy back MEGA. The first buyback on May 7 used net USDm issuer rewards through April; USDm supply sat around $480M, up hard from launch, while future buybacks are supposed to become programmatic and onchain. Nearly 80% of USDm supply sits in Aave V3, with 89% of those deposits concentrated in one entity, Ethena. Elegant flywheel, crowded input.

DRIFT — DRIFT is a recovery-and-relaunch trade, not easy to manage in the short term. Even though they published a user recovery framework for the April 1 DPRK-linked exploit, with total user losses near $295.4M, recovery tokens mapped to verified losses, a pool seeded around $3.8M, Tether support up to $127.5M, partner contributions, and a planned Q2 relaunch, competition is stronger than ever, now facing not just Hyperliquid, but also Jupiter and most recently JTO’s JTX on Solana.

JTO — JTO got fresh institutional and token-accrual headlines last week in addition to JTX’s big product announcement. The Jito Foundation and Solana Company announced APAC validator infrastructure across Hong Kong, Singapore, Japan, and South Korea, with those validators running Jito BAM and co-developing JitoSOL staking/yield products for institutions. Separately, JTX was announced as a self-custodial trading venue combining spot, perps, and prediction markets, with 80% of platform revenue slated to accrue to JTO holders.

JUP — One of the protocols with the largest vertical integration footprint, Jupiter entered the tokenized equities game with Securitize and Jump Trading Group announcing fully onchain regulated trading for tokenized equities on Solana. Securitize brings the regulated broker-dealer/ATS/transfer-agent stack, Jump supplies liquidity via PropAMM, and Jupiter is the user-facing distribution layer.

VVV — Private AI demand, revenue-linked burns, scheduled emissions cuts, and now robotics adjacency through StrikeRobot. Every new subscription triggers a VVV buy-and-burn ($2 Pro, $5 Pro+, $10 Max) on top of discretionary burns, which in total exceed 33.7M VVV, or roughly 42.9% of the original 100M supply. The May 1 emissions cut from 6M to 5M per year is live, with further cuts planned in June and July.

After such a massive move, up 700% on the year and 2x since May, it’s worth keeping the long-term outlook in mind. Last month burned $203K worth of VVV, or $2.4M annualized, which only offsets ~5% of the ~$45M worth of VVV emitted yearly.

POD — Trading as second-order VVV beta, Dolphin/dphn.ai provides the Dolphin Mistral 24B Venice Edition model, which is Venice’s default “Uncensored” model. To kick off the week, POD spiked above $170M market cap before cooling toward ~$130M, with 24h volume around $6.3M for a >80% move. That shows how there are still reflexive, thin, and attention-led opportunities in the market.

CHIP — Consolidating after a strong launch in a market environment dominated by negative sentiment, USD.ai’s main developer, Permian Labs, has started protocol buybacks, repurchasing 338,806,273 CHIP (3.39% of total supply) from investor allocations. Including a prior airdrop-linked buyback, cumulative repurchases reached 3.71% of total supply.

ENS — ENSv2 detailed the move from one flat global registry to a hierarchy of linked registries, so each label can carry its own owner, resolver, subregistry, and rules. That makes ENS more useful for subnames, brands, agents, chain identities, and app-native namespaces without fragmenting into .whatever silos. ENS also posted a May ecosystem snapshot claiming 35M ENS-powered names, which supports the distribution angle. The flaw is token capture with about $2.2M annualized revenue and no obvious holder revenue pipe.

GNO — Last week’s controversial vote on GIP-150 failed. The proposal would have let GNO holders burn tokens for a pro-rata share of the GnosisDAO treasury, with liquid assets distributed at NAV and illiquid/offchain assets represented through a gLTD-CLAIM token. The final voting ended May 12 with 71.9% against, 27.9% for, and 802% quorum. The RFV trade is gone but the governance scar remains. GNO holders now know the treasury discount can be surfaced, measured, and politicized, while the operating side showed it can defend the treasury when activists force the issue.

Equities

SPX — Pressing records around the 7,450–7,470 zone while macro keeps throwing bricks: hot CPI/PPI, retail sales up 0.5% with inflation doing part of the work, and Fed hike odds rising into year-end. Price still wins, but breadth is ugly; Schwab flagged an index rally where more than 300 S&P names were down and equal-weight kept slipping. Respect trend above 7,400. Below 7,300, the AI-led bid starts looking thin.

NDQ — NDQ/QQQ remains the cleanest expression of the AI bid, with QQQ near $721 and Nvidia back in command after H200 China clearance pushed its valuation toward $5.6T. Cisco’s +14% move on hyperscaler orders tells the same story: buyers still pay for AI infrastructure, even with rates pushing back. Chase risk is high; pullbacks above QQQ $714–705 are buyable. Below $700, momentum gets less forgiving.

CRCL — Published Q1 earnings and announced a $222M ARC token sale at a $3B FDV, with a16z crypto, BlackRock, Apollo, ICE, Standard Chartered Ventures, Bullish and others in the book. Arc is built for stablecoin gas, deterministic sub-second settlement, compliance-ready privacy, and direct Circle-stack rails across USDC, CCTP, Gateway, and on/offramps. The ARC token will be inflationary, with the rate of inflation continually declining. These emissions will be allocated as validator rewards. The token is initially distributed by allocating 15% to the treasury, 25% to Circle, and 60% to app builders.

Take Apollo and BlackRock as an example, both of which own big chunks of CRCL. ARC barely matters to them, but any token value shows up as a stronger balance sheet and better financials for Circle. That’s what moves the stock, and that’s where their upside is.

COIN — COIN is the liquid equity proxy for USDC + Base + x402 + “everything exchange” optionality. After last week’s layoffs (~14% of headcount), Coinbase’s thesis hinges on more operational efficiency driving their transition from a crypto exchange into an operating system for finance, crypto, and AI. This is supported by USDC distribution, Base settlement, derivatives, prediction markets, and agent payments. Despite missing earnings with EPS -$1.49 vs $0.04 (miss by 3825%), subscription/services reached 44% of net revenue, derivatives volume was up 169% YoY, retail derivatives are annualizing $200M+, and prediction markets hit $100M+ annualized revenue in March.

NVDA — The market’s favorite toll collector despite such a high market cap bar. Nvidia will invest up to $2.1B in IREN as part of a deal to deploy up to 5GW of AI data-center infrastructure. Earnings land on May 20, 2026. This week it reached $5.5T in market cap as Jensen Huang was added at the last minute to be boarded alongside Trump and other CEOs on their sales trip to China. That’s larger than the GDP of all countries except the U.S. and China.

GOOGL — Google is the best positioned hyperscaler now that AI spend is starting to show up in operating data. Q1 revenue rose 22% to $109.9B, Google Cloud revenue grew 63% to $20.0B, operating margin expanded to 36.1%, and the company booked $37.7B of other income, mostly unrealized gains on non-marketable equity securities. Unlike its competitors in AI, Alphabet has a strong track record in Cloud and sufficient capital to back its AI thesis without staying underwater for long periods of time.

META — Polar opposite of Google. Capex is colliding with investor patience. Q1 revenue rose 33% y/y to $56.31B, ad impressions rose 19%, average price per ad rose 12%, and free cash flow was still $12.39B. Then management lifted 2026 capex guidance to $125–145B, citing higher component pricing and future data-center capacity. Overall, a stock that is more fragile than the headline beats suggest.

Anthropic — The Anthropic bid had gone vertical onchain trading on Jupiter perps pre-IPO in the midst of Anthropic weighing a summer raise that could push valuation near $1T. Its last official round was only February 12 at $380B, with $30B raised and $14B of run‑rate revenue disclosed by the company. Claude’s run‑rate revenue has passed $30B, up from roughly $9B at year-end 2025, while customers spending more than $1M annualized doubled to over 1,000 in less than two months. On May 6 Anthropic added a SpaceX compute deal for 300MW+ and 220,000+ NVIDIA GPUs.

The vertical move showed where trapped demand could actually trade, which includes crypto venues like PreStocks/Jupiter-style tokenized exposure, pre‑IPO perps, Hiive, Forge, and other secondary wrappers. However, things went south right after the PreStocks Anthropic token hit $1,409.20 on May 11. Price snapped back near $800 on May 13 as Anthropic updated its stock-sale warning: unapproved transfers, SPVs, forward contracts, tokenized securities, and new offerings from named platforms including Forge and Hiive are void and give buyers no shareholder rights.

This shouldn’t have caught anyone by surprise. PreStocks itself says its products provide economic exposure only, with no ownership, voting, dividend, information, or legal rights, and no guaranteed liquidity. Although the read-through for tokenized OpenAI, SpaceX, xAI, Databricks, Anduril, and similar pre‑IPO instruments is that the market can still trade synthetic exposure while actual issuer recognition + legal enforceability + liquidity depth will become increasingly important.

SpaceX — The FCC approved EchoStar’s $40B spectrum sale on May 12; SpaceX gets 65 MHz for $17B to support Starlink direct-to-device, with waivers for terrestrial, space-based, and hybrid network use. At the same time, SpaceX is scouting U.S. and overseas spaceport sites because Starship’s economic case depends on reuse and launch frequency. This comes to show that SpaceX is becoming telecom infrastructure plus transport infrastructure, with the IPO story increasingly tied to regulated network rights and launch cadence.

OpenAI — Doubling down on their enterprise strategy to challenge Anthropic, also drawing inspiration from Palantir’s operating model, OpenAI launched the OpenAI Deployment Company, majority owned and controlled by OpenAI, with more than $4B of initial investment and Tomoro’s ~150 forward-deployed engineers coming in to help businesses build around intelligence. This week they also opened GPT‑5.5‑Cyber access to Deutsche Telekom, BBVA, Telefónica, Sophos, Scalable Capital, and other European companies under Trusted Access for Cyber.

Cerebras — The AI chipmaker, famous for its fast inference speeding up LLMs, debuted on the Nasdaq post-IPO (priced at $185) opening at $350. For context, the May roadshow started at 28M shares at $115–$125, then demand north of 20x pushed it to 30M shares at $150–$160, and final pricing landed at $185, raising $5.55B at a $56.4B fully diluted valuation (roughly 111x 2025 revenue of $510M). Arm and SoftBank tried to buy it in the final stretch, but management declined on the expectation that public markets would pay more than the strategic M&A.

Conclusion

The market keeps rewarding assets tied to visible cash flow and punishing anything that asks for patience without a buyer. AI still has more tailwinds than crypto. Its spending shows up in orders, revenue, and index weight before anyone forces the productivity question. Crypto has to work harder: BTC can lean on structural demand, HYPE can point to revenue, tokenized finance can point to rails, but most beta still needs someone else to arrive. The rule this week is to respect the rally without romanticizing it: buy proof, fade crowded belief, and cut anything that needs perfect rates, perfect oil, and perfect positioning to survive.

Alea Research · 16 May 2026 · 22 min

Back to top