TMTB: BofA/William Blair Conference Summaries - NVDA, NOW, BKNG, MCHP, ADSK, CSCO, TWLO

NVIDIA — Speaker: Ian Buck (VP Accelerated Computing)

Sounded bullish although presentation was pretty technical…Buck painted an even larger inference horizon and repeatedly positioned NVIDIA as the leader in that world and had some good points re: Deepseek

— DeepSeek catalyst & reasoning token explosion
“The January DeepSeek moment democratized world-class reasoning—a 671 B model that talks to NVLink and GPUDirect, and its openness ‘barn-doored reasoning for everyone.’…Because each 100-token answer now calls for ~1 000 tokens of thought, total tokens per user jump 13–20×, potentially expanding the inference revenue pool twenty-fold.”

— GB200 built for multi-GPU reasoning
“That surge ‘arrived at the perfect time for GB200,’ whose NVLink-fed multi-node fabric is purpose-built for long-context inference, turning what used to be a training-only chip into an inference workhorse.”

— Inference eclipsing training in profit pool
“Saturation fears miss the point: with richer reasoning, inference becomes the larger profit pool, re-orienting cloud cap-ex road-maps toward sustained accelerator demand.”

— Open-source acceleration favors NVIDIA stack
“By releasing code and tricks, DeepSeek guarantees the next wave of open models will be NVLink-optimized from day one, reinforcing our platform advantage.”

— MoE + MLA compression breakthrough
“DeepSeek also married Mixture-of-Experts with MLA statistical compression, shrinking memory while preserving accuracy—tricks once limited to image gen are now mainstream for LLMs, bending the cost curve by engineering chops, not cap-ex alone.”

---

ServiceNow — Speaker: Chief Customer Officer Chris Bedi

Bedi rattled off tangible AI wins and conveyed a sense of urgency and confidence in near-term monetization.

— AI-first transformations in marquee accounts
“Stellantis is leveraging Now to re-architect shared-service cost structure, dealer ops and supply-chain AI, while AstraZeneca runs agentic workflows in drug discovery and regulatory…these ‘end-to-end AI-first stories’ showcase real value, not pilots.”

— Monetization via Pro Plus & out-of-box agents
“Boards say ‘AI is no longer optional,’ and Pro Plus plus turnkey agents are driving rapid attach; analyst-day made clear monetization is scaling now, not three years out.”

— Federal & vertical momentum
“Federal revenue grew 30 % y/y even amid head-count cuts, and verticalized go-to-market hits 90 % penetration where deployed, opening new LOB budgets beyond tech.”

— Now Next AI sprints & reusable IP
“Within 30 days we ran 40+ Fortune-500 AI design sessions, forward-deploying engineers and creating reusable IP that fuels repeatable workload sales.”

— G2K penetration & white-space
“We’re 60 % penetrated in the Global 2000, leaving a 40 % international white-space we’re actively targeting through vertical plays and strategic account programs.”

---

Twilio – Speaker: CEO Jeff Lawson

Lawson sounded confident that tighter focus, AI-powered efficiency, and high-margin add-ons will keep the margin ramp intact while nurturing a smaller set of breakthrough products.

— Margin expansion & guidance philosophy
“We have already pushed operating margin into the high-teens—an enormous swing from ‘losing money and burning cash’ just two years ago…automation and AI are expected to generate further OpEx leverage, so the upside to guidance ‘comes from doing better every quarter’ and gives us ‘additional margin-expansion capability’ over the medium term.”

— Gross-margin drivers: Verify, Voice & unified API
“Newer software-centric products such as Verify and Fraud Guard will layer ‘software-like margins’ on top of Messaging, while Voice’s attractive economics and the ConversationRelay unified API should ‘push gross margins higher over time.’”

— AI strategy: partner, don’t build an LLM
“We will not build our own LLM—budget alone makes that impractical…instead the platform ‘plugs into best-of-breed models with a click,’ and an Azure bake-off proved the partner’s AI stack ‘extremely compelling,’ letting customers bring voice-centric AI use cases to life quickly.”

— Product-development playbook (fewer, bigger bets)
“The new approach is to focus on ‘three to five high-efficacy growth bets’ rather than ten science projects…we invest small and early, measure traction, then allocate more resources only when signals are clear, which keeps burn low and hits the market faster.”

— Platform consolidation & easy API access
“All products are now integrated into a single platform, exposed through simple APIs, so developers can tap any channel, enrich interactions with context, and store data for full lifetime value—all ‘without hopping among silos.’”

— ConversationRelay traction
“ConversationRelay is live and “has seen fantastic uptake”—customers are ‘super-excited’ because one unified API works across multiple channels and carries the richer margin profile we just discussed.”

— Automation, AI & OpEx leverage
“Management keeps repeating that AI-driven automation is ‘largely untapped’ inside Twilio…the same tooling that boosts customer outcomes is also set to ‘drive operational-expenditure leverage and short-term margin expansion.’”

— Capital allocation & buybacks
“We’ve been ‘very focused on buybacks recently,’ signaling confidence in long-term growth while keeping optionality for selective M&A.”

— Workforce rightsizing & self-funded investment
“After difficult layoffs the team is ‘the right size’; head-count additions aren’t needed to pursue growth bets, which can be ‘self-funded by efficiency gains.’”

— Consumer engagement & macro backdrop
“Our data show that when brands create more intimate, value-adding communications—pushed promotions via SMS, richer voice interactions—consumers ‘don’t mind more messages’ and in fact spend more, proving the demand side of the engagement thesis even in a choppy macro environment.”

---

Booking Holdings — Speaker: CFO Ewout Steenbergen